JP4188200B2 - Plant-wide optimum process controller - Google Patents

Description

  The present invention relates to a plant-wide optimum process control apparatus for process systems such as a sewage treatment process, a wastewater treatment process, a water purification process, and a chemical process.

  Conventionally, PID control and sequence control, which are types of classical control methods, have been widely used as control methods for industrial processes such as water and sewage water quantity processes and water quality processes. The PID control method and the sequence control method are balanced control methods from the viewpoints of reliability, control performance, ease of design and tuning, and are still the most standard control methods in the industry. Used to control more than 100% of industrial processes.

  These basic control methods are usually constructed as a so-called 1-input 1-output system (usually called “one-loop controller” as a control device name) that controls one manipulated variable for one purpose. However, such a control device is no longer sufficient to meet various requirements such as the following.

  For example, in the field of sewage treatment, environmental considerations have become more important in recent years, and the regulation of discharge into rivers such as nitrogen and phosphorus has become mandatory due to the conventional requirement for organic matter removal only. There is a growing need to control these multiple water qualities simultaneously. In addition, there is a demand not only for simple control performance, but also for cost reduction and energy saving, as in the trade-off seen in improving the quality of treated water and the demand for reducing the amount of chemicals injected in the water and sewage treatment process. The purpose of control is diversifying. In order to meet such demands, various advanced controls have recently been used in addition to conventional PID control and sequence control.

  Advanced control is roughly classified into two approaches, rule-based control and model-based control, and these are used properly according to the purpose. If there is a model that can appropriately represent the behavior of the target process, the model-based control can potentially obtain better control performance than the rule-based control (for example, Patent Documents). 1 and Patent Document 2). Therefore, if there is a model that can simulate the behavior of the target process, the model-based control has the potential to perform better control.

In the field of control of chemical processes and water and sewage processes, so-called process control, there is a strong demand for reducing the operating costs of chemicals and electric power to be controlled at the same time as the demand for control performance. Model predictive control, which is one of model-based controls, has been developed (see, for example, Non-Patent Document 1). Model predictive control is a control method that has been put to practical use in the industry by DMC (Dynamic Matrix Control), which is independent from Shell Oil, from the late 1970s to the early 1980s. Many methods are used. However, in the DMC, for example, a control law (control method) is configured as a process model based on a “relatively simple process model” such as a “step response model” well known in control engineering. If it becomes complicated, it cannot cope. For example, in a complex process with strong nonlinearity such as a sewage treatment water quality process, a model that simulates the behavior of the water quality process, which is the model (usually called a “simulator”), is also used for higher-order nonlinearity. It is expressed using complex mathematical expressions centered on differential equations, resulting in a “complex process model”. It has been difficult to directly apply the conventional model predictive control method to such a “complex process model”. Therefore, in many cases, a “simple process model” is derived from the “complex process model” and a control law is constructed based on the “simple process model”. For example, in the field of sewage treatment, the International Association on Water Quality (IAWQ) has established activated sludge models ASM1 to ASM3 (Activated Sludge Model No. 1 to Activated Sludge Model No.) as models for nitrogen and phosphorus removal processes. 3) is published, but these models are expressed by higher-order (11th to 16th order) nonlinear differential equations, and it is difficult to establish a control law using them directly. Usually, control based on an approximate model is performed (see, for example, Non-Patent Documents 2, 3, 4, and 5). The reason why such a “simple process model” is used for control is not only because it is practically easy to handle, but also because of the robustness that is a feature of feedback control. In general, feedback control is known to be able to perform robust control over the gap that exists between the process model and the actual process if the control law is appropriately configured. In many cases, a “complex process model” is not always necessary.

  On the other hand, the “simple process model” does not accurately represent the target process, and a “complex process model” that can express the process in more detail and accuracy is used to construct a direct control law. There is also a deep thought that it should be. If a direct control law can be constructed using a “complex process model”, it is possible not only to establish a control method with better control performance, but also to facilitate coordination between process control and process operation management and process monitoring. Furthermore, it becomes easier to cooperate with process abnormality diagnosis, etc., and there is a possibility that an optimal operation control system that is plant-wide can be established. However, as described above, it is generally difficult to construct a model-based control based on a “complex process model”.

  As described above, “model-based control based on a simple model” and “model-based control based on a complex model” have their own merits and demerits, but in reality, they are binary classifications of “simple” or “complex”. Instead, it is important to close the gap between them and establish a control system using a “model with a degree of complexity necessary to achieve the objective”. From this point of view, in the invention described in Patent Document 1, a simple process model for control is identified from a complex process model, and a control law is derived through this process model. In the invention described in Patent Document 2, a “direct process simulator applied non-linear control device” that directly derives a control law by using a response (output) of a complicated process model is proposed. Has proposed.

As shown in FIG. 13, this proposal is a potential control method using a process simulator that can be used in various actual industrial processes (2000) in the future, such as water purification processes, sewage treatment processes, and petrochemical processes. However, consistency with the PID control method (2010) generally used in process control at present is not necessarily good. It has already been mentioned that the PID control system is used in over 80% of the control of industrial processes. The control A according to Patent Document 1 and Patent Document 2 is to fill a gap between “model-based control based on a simple process model (2022)” and “model-based control based on a complex process model (2021)”. In this case, a “control method applicable to an actual process (2000)” is provided, but “actually used control (2010)” and “model-based control (2020)” are provided. It does not provide a solution to fill the gap. Therefore, it is not possible to introduce a new control method based on a process model that seamlessly extends “currently used control (2010)”, and it is necessary to construct a new control method from scratch. However, from a practical point of view, if the “currently used controls” can be expanded in a natural way, various know-how and knowledge accumulated through conventional experience can be inherited. It is easy to accept, and it is considered that a new control method can be spread to the industry relatively easily.
JP 2001-249705 A JP 2002-373002 A J. Richalet, "Industrial Implementation of Predictive Control", Preprint of Japan Batch Forum, (2002) H. Brouwer et al, "Feedforward Control of Nitrification by Manipulating the Aerobic Volume in Activated Sludge Plants", Wat. Sci. Tech. Vol. 38, No. 3, PP.245-254, (1998) HJS Lukasse et al, "Adaptive Receding Horizon Optimal Control of N-Removing Activated Sludge Processes", Proc. Of Med. Fac. Landbouww. Univ. Gent pp. 1665-1672, (1997) SR Weijers et al, "Control Strategies for Nitrogen Removal Plants and MPC Applied to A Pre-Denitrification Plant", Proc. Of Med. Fac. Landbouww. Univ. Gent pp. 2435-2443, (1995) Yamanaka, Nagaiwa, Tsutsumi, Nagamori, Hatsuka "Sewage treatment plant denitrification / dephosphorization water quality control using model predictive control", Proceedings of the 8th Control Technology Symposium, pp. 459-462, (2000) Yamanaka, Ohara, Motoki, Tsutsumi, Hatsuka "Examination of advanced treatment water quality control considering operational costs", Proceedings of the 39th Sewerage Research Conference, pp. 233-235, (2002) IAWQ Task Group on Mathematical Modeling for Design and Operation of Biological Wastewater Treatment Processes, "Activated Sludge Models ASM1, ASM2, ASM2d and ASM3", IAWQ Scientific Technical Report No. 9, (2000) J. Kappeler and W. Gujar, "Estimation of Kinetic Parameters of Heterotrophic Biomass Under Aerobic Conditions and Characterization of Wastewater for Activated Sludge Modeling", Wat. Sci. Tech. Vol. 25, No. 6, pp. 125-139, ( (1992) U. Jeppson, "Modeling Aspects of Wastewater Treatment Processeso", PhD thesis. Lund Institute of Technology ", (1996)

  An object of the present invention is to provide an optimal control device that can seamlessly combine “currently used control” and “model-based control using a process model” based on the above concept. If this is described with reference to FIG. 13, the present invention can seamlessly combine “currently used control (2010)” and “model-based control using a process model (2020)”. It can be said that the object is to provide the optimal control B.

  In order to achieve this object, the present invention proposes a control method that appropriately combines the following three requirements for process control.

(1) Dynamic compensation of process dynamics
(2) Compensation consistent with conservation laws such as material balance and energy balance (a kind of static compensation)
(3) Optimization in consideration of cost and a plurality of control purposes simultaneously (Process optimization).

  PID control, which is a representative example of “currently used control”, considers only the request (1). In addition, indirect process simulator application control (model prediction control) of Patent Document 1 considers requirements (1) and (3) at the same time, and (2) is indirectly considered. Various control methods developed in the control theory are usually built around the requirements (1) and (3), and include “robustness”, “adaptivity”, “nonlinearity”. Theories have been developed to fill in gaps with actual processes by adding concepts such as “sex”, “unsteadiness”, and “distribution”. In the control theory, the requirement (2) can be indirectly considered by using concepts such as “dissipative”, “passivity”, and “lossless”. However, control methods using these concepts are mainly applied to mechanical and electrical control targets, and are not often used for process control targets. On the other hand, in model predictive control often used in process control, generally, a control method has been developed taking into account requirements (1) and (3). An example that actively considers requirement (2) is “Model Predictive Control Based on First Principle (Physical Law)” in Non-Patent Document 1. The author of this document points out the importance of "control based on the first principle (physical law)", but it has only been applied to a relatively simple process of bilinear type and low order.

  In the present invention, all of the requests (1), (2), and (3) are considered. At this time, the controller that considers the requests (1) to (3) at the same time is not configured, but the entire controller is configured. Is constructed hierarchically in a form including a conventional controller, and each element of the controller is appropriately assigned each of the requests (1) to (3). The basic idea of the present invention is as follows.

  The dynamic compensation of the request (1) process is basically left to conventional methods such as PID control. This is because it is not always necessary to use an advanced control method for dynamic compensation, and a conventional control method such as PID control is often sufficient. In particular, unlike electrical and mechanical controlled objects, process-related controlled objects are usually more important for steady-state characteristics with static compensation than for transient characteristics with dynamic compensation. There is often no need to spend so much effort.

  Concepts such as the material balance and energy balance of requirement (2) are expressed as algebraic equation constraints as a static process model. This is because the concept of these balances is one of the laws of conservation, which is one of the laws of physics, but when considering the steady state, the conservation laws can be expressed by algebraic equations rather than differential equations or partial differential equations. is there. Since it is often pointed out that it is important to consider the laws of physics (first principles), “dynamic process models = simulators” are often used as process models based on the first principles. If only a conservation law in a steady state is required as a principle, an algebraic equation that is one of the expressions of “static process model” is sufficient. In fact, listening to the opinions of people who point out the importance of laws of physics, in fact, often refers to the conservation law in steady state, which is a special case. Therefore, if only the conservation law in the steady state is used as a problem, a constraint by an algebraic equation may be used.

  The optimization of requirement (3) is not “dynamic optimization with dynamic constraints (differential equation constraints)” as in conventional model predictive control, but static / dynamics with static constraints (algebraic equation constraints). By “optimization”. This is made possible by expressing the process model with a “static process model” instead of a “dynamic process model”.

  In the present invention, in accordance with this idea, the conventional control method, such as a PID control unit, is arranged in a lower system, that is, a place close to the target process, such as the one-loop controller described above, and the upper system, that is, the target process. The first object is to provide a process optimization control device having an optimization device that is incorporated in an EWS (engineering workstation) or the like that is far from the eye and has a conservation law constraint expressed by an algebraic equation as a static process model. 1 purpose.

  It is a second object of the present invention to provide a plant-wide optimum control apparatus including an abnormality diagnosis apparatus that performs abnormality diagnosis when the optimum control cannot be achieved due to process abnormality.

  By the way, all of these technologies are strongly aimed at realization in industrial processes, but another important way of thinking is necessary in realizing process control. It is the concept of “robustness”. “Robust” is a word that means “robust”, and in the field of control theory, robust control theory has been developed mainly for the purpose of model uncertainty and robustness against disturbances. Even in public / industrial processes such as water and sewage processes, which are target processes of the present invention, it is difficult to actually realize a process control system in an actual plant unless robustness is taken into consideration. However, until now, there has been hardly any optimal control method for the whole plant that specifically considers robustness. Accordingly, a third object of the present invention is to provide robustness to the optimal control method for the entire plant.

  Here, there are various types of robustness. Typical examples include robustness to process model parameter fluctuations, robustness to the effects of higher-order dynamics of the process, robustness to disturbances in the process, and such robustness for the third purpose. It is not intended to provide an optimal process control device that takes into account the above. The plant-wide optimum process control apparatus provided for the third purpose is a process control apparatus that mainly considers robustness to avoid failure of process control due to sensor abnormality.

  In particular, in processes where there are many contaminants, such as sewage treatment processes, and sensor abnormalities are likely to occur, robustness against sensor abnormalities is required when actually operating a process control system rather than robustness as described above. There are many cases. Therefore, in the invention according to the third object, even when some abnormality occurs in the sensor, the control can be performed while avoiding the failure of the process.

  The plant-wide optimum process control apparatus according to the first aspect of the present invention includes at least one process sensor capable of measuring the state of the target process, at least one actuator capable of changing the state of the target process, and a target process. In a plant-wide optimal process control device for controlling a process having an external input, the actuator is operated based on the process measurement value obtained by the process sensor, and the local variation for compensating the dynamic variation of the process to have a desired characteristic A control unit, a constraint condition setting unit that sets at least a static constraint condition regarding the conservation law of the storage amount such as the amount of substance and energy of the external input of the process, and an evaluation function that evaluates the operating cost and multiple control performance Optimal evaluation function setting part to be set and constraints An optimal target value calculation unit that inputs the constraint condition set by the condition setting unit and supplies the optimal target value of the local control unit based on the evaluation function set by the optimal evaluation function setting unit It is characterized by.

  According to the present invention, a control system that optimizes a process within a range that satisfies some conservation law that the process must essentially satisfy without constituting various control methods conventionally used as a local control unit is configured. can do. Furthermore, the local control unit of the lower system and the optimum target value calculation part of the upper system can be operated independently. For example, if you want to improve the dynamic characteristics of the process, such as disturbance removal characteristics, adjust only the local control unit. In addition, if you want to change the optimization criteria, you only need to change only the higher-level optimal target value calculation unit. Can be adjusted. Furthermore, because the process imposes a conservation law that the process should essentially satisfy as a constraint condition, the target value that cannot be reached in principle by optimization without such a conservation law constraint condition is set. May be destabilized, but according to the present invention, in principle, it can be achieved unless the process becomes abnormal due to, for example, certain reactions being inhibited. Target value can be set. Therefore, the plant can be controlled stably.

  The invention according to claim 2 is the plant-wide optimum process control device according to claim 1, wherein the local control unit given the optimum target value calculated by the optimum target value calculation unit controls the process, and as a result, from the process sensor A process abnormality notification device is provided for notifying a process abnormality when the derived controlled quantity cannot achieve the optimum target value.

  According to the present invention, in addition to the effect of the first aspect, by notifying the plant operator or the like of an abnormal phenomenon that the target value that should be reachable due to the constraint condition of the conservation law cannot be reached, the process becomes inefficient. Stabilization can be prevented.

  The invention according to claim 3 is the plant-wide optimum process control device according to claim 1, wherein the local control unit given the optimum target value calculated by the optimum target value calculation unit controls the process, and as a result, from the process sensor. Equipped with a process abnormality diagnosis device that, when the derived controlled quantity cannot achieve the optimal target value, determines that the process is abnormal, automatically investigates the cause of the abnormality, and reports the investigation results to the process manager It is characterized by that.

  According to the present invention, in addition to the effect of claim 1, in addition to notifying the operator of the process abnormality as in claim 2, not only the operator but also the cause can be notified, and the process is quickly restored to the normal state. Can help.

  According to a fourth aspect of the present invention, in the plant-wide optimum process control device according to the first aspect, the local control unit given the optimum target value calculated by the optimum target value calculating unit controls the process. Incorporates a mechanism that, when the derived controlled quantity cannot achieve the optimum target value, determines that the process is abnormal, automatically investigates the cause of the abnormality, and automatically restores the process based on the investigation result It is characterized by having a control device for abnormal process.

  According to the present invention, not only the optimum control at the normal time that does not cause a process abnormality but also the control at the time of abnormality even when the process becomes abnormal, the process robustness is remarkably improved. Can be improved.

  The invention according to claim 5 is the plant-wide optimum process control apparatus according to claim 1, further comprising a manual target value calculation unit that can be switched between the optimum target value calculation unit and the optimal target value calculation unit.

  According to the present invention, when the target value obtained by the optimum calculation cannot be satisfied, the process manager and the operator can intervene, and the plant can be freely operated without changing the performance of the local control unit. it can.

  The invention according to claim 6 is the plant-wide optimum process control device according to claim 1, wherein the constrained nonlinear programming is applied to the optimum target value calculation unit.

  According to the present invention, a reliable optimum target value can be set by performing an optimum target value calculation by a nonlinear programming method known as a reliable optimization calculation method in various fields. It is possible to construct an optimal control system for the plant wide.

  The invention according to claim 7 is the plant-wide optimum process control apparatus according to claim 1, wherein a semi-optimization method based on metaheuristics is applied to the optimum target value calculation unit.

  According to the present invention, even when the conservation law is expressed by a constraint condition based on a non-linear algebraic equation or the evaluation function for optimality is a non-linear evaluation function, it is possible to avoid falling into a local optimal value as much as possible. Since an answer close to a global optimum solution can be obtained, a reliable optimum or suboptimal target value can be set. This makes it possible to construct a control system that is optimal for the plant wide.

  The invention according to claim 8 is the plant-wide optimum process control apparatus according to claim 1, wherein static constraint conditions relating to conservation laws such as material balance and energy balance of the constraint condition setting unit are predetermined according to changes in external input. And the calculation in the optimum target value calculation unit is updated in synchronization with this.

  According to the present invention, static optimization can be replaced with pseudo-dynamic optimization in accordance with external change conditions such as changes in the quality of treated sewage water and changes in the quality of intake water of purified water. It can be evaluated more precisely.

  The invention according to claim 9 is characterized in that an external input related to a static constraint condition concerning a conservation law such as a material balance and an energy balance of the constraint condition setting unit is constructed from only measurable variables.

  According to the present invention, since the optimization calculation can be performed using only the measurable variables of the external input, the plant-wide optimal process control device is immediately mounted without changing the basic idea of the present invention at all. be able to.

  The invention according to claim 10 is the plant-wide optimum process control device according to claim 1, and can measure external inputs related to static constraint conditions related to conservation laws such as material balance and energy balance of the constraint condition setting unit. And an unmeasurable variable, and has an estimation means for estimating the value of the unmeasurable variable.

  According to the present invention, by estimating the value of an unmeasurable variable of an external input, the unmeasurable variable can be introduced into the constraint condition, and as a result, the process can be optimized more precisely.

  The invention according to claim 11 is the plant-wide optimum process control device according to claim 1, wherein the parameter of the control law of the local control unit is changed according to the optimum target value supplied from the optimum target value calculation unit. Features.

  According to the present invention, at an operating point near a given target value, it is possible to provide optimal performance for local control, such as target value tracking characteristics, disturbance suppression characteristics, robustness, etc. In addition to the optimization of the transient characteristics, the transient characteristics can be optimized.

  The invention according to claim 12 is the plant-wide optimum process control apparatus according to claim 1, wherein the local control unit has a control law using a dynamic model of the process.

  According to the present invention, a process model is used even for a process having strong nonlinearity such as a sewage water quality process, and a strict linearization method, input / output linearization method, backstepping method, passivity By applying a control method based on the above, it is possible to construct a control system having robustness with respect to any target value.

  The invention according to claim 13 is the plant-wide optimum process control device according to claim 1, further comprising a display device for displaying an output result of the optimum target value setting unit, wherein the local control unit is a process manager / operator. Is manually operated.

  According to the present invention, even in a plant where the automatic control is not actually used and the operator is operating manually, the operator can be advised on the optimum operation method, and as a result, the optimum Process operation can be made possible.

  The invention according to claim 14 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the optimum evaluation function setting unit relates to the operation of the discharged water quality cost converted into monetary amount and the operation. It is characterized by using a cost evaluation function that simultaneously considers operating costs.

  According to the present invention, an optimum sewage treatment process control device can be constructed in consideration of the trade-off between improvement of the discharged water quality of the sewage treatment process and reduction in operation cost.

  The invention according to claim 15 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and as a constraint condition, a variable that can be converted into a COD concentration with respect to inflow sewage as an external input, Variables that can be converted to TOC, variables that can be converted to BOD, variables that can be converted to ammonia concentration, variables that can be converted to nitric acid concentration, variables that can be converted to Kjeldahl nitrogen concentration, and conversion to total nitrogen concentration Possible variables, variables convertible to phosphate concentration, variables convertible to total phosphorus concentration, variables convertible to MLSS concentration, variables convertible to SS concentration, variables convertible to pH, The mass conservation law is described by an algebraic equation using at least one of a variable that can be converted into ORP and a variable that can be converted into alkalinity.

  According to this invention, in the sewage treatment process, an optimal control system can be constructed using only measurable variables. In particular, an optimal control system for the purpose of removing nitrogen and phosphorus can be constructed.

  The invention according to claim 16 is the plant-wide optimum sewage treatment process control device according to claim 15, wherein, in addition to the variable of the constraint condition, as an external input, a plurality of elements of organic matter measured by COD and measured by MLSS Using the variables that represent the multiple elements of sludge containing microorganisms, the conservation law of the substance amount is described by an algebraic equation, the multiple elements of organic matter measured by COD, and the multiple elements of sludge measured by MLSS And an estimation means for estimating.

  According to the present invention, for example, acetic acid-based organic matter having an important influence on phosphorus removal, individual behaviors of a plurality of types of microorganisms involved in removal of nitrogen, phosphorus and the like can be incorporated into a conservation law as a constraint condition, A sewage treatment optimum control system can be constructed more strictly.

  The invention according to claim 17 is the plant-wide optimum process control apparatus according to claim 1, wherein the process is a sewage treatment process, and the local control unit uses a process sensor to determine the amount of aeration air that is one of the objects to be controlled by the actuator. Control is performed using the detection output of at least one of the DO meter, ORP meter, ammonia meter, nitric acid meter, and TN meter.

  According to the present invention, optimal aeration control can be performed, and not only sewage treatment process control that optimizes organic substance removal and nitrogen removal can be realized, but also the power cost associated with aeration can be reduced.

  The invention according to claim 18 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the local control unit determines the amount of returned sludge, which is one of the objects to be controlled by the actuator. Control using detection output of at least one sensor among ammonia meter, TN meter, TP meter, nitrate meter, phosphate meter, ORP meter, MLSS meter, UV meter, COD meter, BOD meter and TOC meter as sensors It is characterized by doing.

  According to the present invention, it is possible to perform optimum return sludge amount control, not only to implement sewage treatment process control that optimizes organic matter removal, nitrogen removal, and phosphorus removal, but also to reduce power costs associated with pump operation. Can be made.

  The invention according to claim 19 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the local control unit determines an amount of surplus sludge extraction that is one of the objects to be controlled by the actuator. Control is performed using a detection output of at least one of a phosphoric acid meter, a nitric acid meter, an ammonia meter, a TP meter, a TN meter, and an MLSS meter as a process sensor.

  According to the present invention, it is possible to perform optimum control of the amount of excess sludge withdrawing, and not only can implement sewage treatment process control optimized for organic matter removal, nitrogen removal and phosphorus removal, but also reduce the power cost associated with pump operation. Can be

  The invention according to claim 20 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the local control unit uses a process sensor to determine a circulation amount that is one of objects to be controlled by the actuator. Control is performed using the detection output of at least one of the nitric acid meter, ORP meter, TOC meter, COD meter, BOD meter, UV meter, and ammonia meter.

  According to the present invention, optimal circulation amount control can be performed, and not only can sewage treatment process control with optimized nitrogen removal be realized, but also the power cost associated with pump operation can be reduced.

  The invention according to claim 21 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the local control unit determines a carbon source input amount that is one of the objects to be controlled by the actuator, Control is performed using the detection output of at least one of a nitric acid meter, a TN meter, an ORP meter, a COD meter, a UV meter, a BOD meter, and a TOC meter as process sensors.

According to the present invention, optimal carbon source input control can be performed, and not only can sewage treatment process control with optimized nitrogen removal be realized, but also chemical costs associated with carbon source input can be reduced. ,
The invention according to claim 22 is the plant-wide optimum process control device according to claim 1, wherein the process is a sewage treatment process, and the local control unit is a carbon source input amount and circulation which is one of the objects to be controlled by the actuator. The amount of organic matter and the amount of nitric acid using the detection output of at least two sensors of the COD meter, UV meter, TOC meter, BOD meter, nitrate meter, TN meter, and ORP meter as process sensors. The ratio is controlled simultaneously so that the ratio becomes a predetermined value.

  According to the present invention, the removal of organic substances and the removal of nitrogen can be optimized at the same time, and further, the chemical cost associated with the carbon source input and the power cost associated with the circulation pump can be reduced.

  The invention according to claim 23 is the plant-wide optimum process control apparatus according to claim 1, wherein the process is a sewage treatment process, and the local control unit determines a flocculant input amount that is one of objects to be controlled by the actuator, Control is performed using a detection output of at least one of a phosphoric acid meter, a total phosphorus meter, and an ORP meter as a process sensor.

  According to the present invention, it is possible not only to carry out optimal flocculant input control, but also to reduce chemical costs associated with flocculant input.

  According to a twenty-fourth aspect of the present invention, in the plant-wide optimum process control apparatus according to the first aspect, the process is a sewage treatment process, and the local control unit is a process sensor that determines the step inflow amount as one of the objects to be controlled by the actuator. Control is performed using the detection output of at least one of the phosphoric acid meter, TP meter, nitric acid meter, TN meter, ORP meter, COD meter, UV meter, BOD meter, and TOC meter.

  According to the present invention, optimal step inflow control can be performed.

  The invention according to claim 25 is capable of measuring at least one actuator capable of changing the state of the target process and several states of the target process, and determining an operation amount of the actuator. In any process having at least one or more process sensors, including an used process sensor, an external input (disturbance) to the target process, and at least one or more external input sensors for measuring the process sensor, the process sensor A process sensor abnormality diagnosing unit for determining whether the external input sensor is in a normal state or an abnormal state, an external input sensor abnormality diagnosing unit for determining whether the external input sensor is in a normal state, or a process measurement obtained by at least one of the process sensors Based on the value, operate at least one of the actuators to The local control unit that compensates for dynamic fluctuations of the external input sensor so as to have desirable characteristics, and the external input sensor abnormality diagnosis unit when the external input sensor is normal. A constraint condition setting unit having at least one static constraint condition regarding a conservation law of a storage amount such as a substance amount and an energy amount of the external input of the process, which can input an external input value analyzed by An optimal evaluation function setting unit that can set an evaluation function that can evaluate an operating cost and a plurality of control performances, and a constraint condition set by the constraint condition setting unit are input and set by the optimal evaluation function setting unit Based on the evaluation function, when the output of the process sensor abnormality diagnosis unit is normal, the optimal target value is output to the local control unit, When the output of the serial process sensor abnormality diagnosis unit is abnormal, characterized in that and a process optimization unit which can output an optimum manipulated variable to the actuator.

  According to the present invention, when no abnormality is recognized in the sensor, it is possible to realize a plant-wide optimum control apparatus mainly using feedback control which is generally robust to process fluctuations. In such a case, although the robustness against the process variation by the feedback control is given up, a control device that determines the optimum operation amount of the plant by the feedforward control can be realized. As a result, even if a sensor abnormality occurs, the process control system does not fail, and the mathematical formula used in the process optimization unit is not significantly different from the actual process. The process can be optimally controlled. Even if the actual process and the mathematical formula are slightly different from each other, the operation amount of the process can be calculated in any situation within a reasonable range. As a result, a plant-wide optimum control device that is robust against sensor failure can be provided.

  The invention according to claim 26 is capable of measuring at least one actuator capable of changing the state of the target process and several states of the target process, and determining an operation amount of the actuator. Any process having at least two or more process sensors, including at least two or more process sensors to be used, and an external input (disturbance) to the target process and at least one or more external input sensors for measuring the same A process sensor abnormality diagnosis unit that determines whether each of the plurality of process sensors is in a normal state or an abnormal state, and an external input sensor abnormality diagnosis unit that determines whether the external input sensor is in a normal state or an abnormal state, Based on any process measurement by at least two of the ranked process sensors A local control unit that operates at least one of the actuators and compensates for dynamic fluctuations in the process to have desirable characteristics; and when the external input sensor is normal in the external input sensor abnormality diagnosis unit, Static constraints on the conservation law of the storage amount such as the amount of substance and energy of the external input of the process, which can input the external input sensor value by the manual analysis etc. in case of abnormality A constraint condition setting unit having a condition as at least one constraint condition, an optimal evaluation function setting unit capable of setting an evaluation function that can evaluate an operating cost and a plurality of control performances, and a constraint set by the constraint condition setting unit Based on the evaluation function set by the optimum evaluation function setting unit, the optimum target value is output to the local control unit. And an optimum target value calculation unit, and a process optimization unit having an optimum operation amount calculation unit that outputs an optimum operation amount to the at least one actuator, and the optimum target value calculation unit Selects one of the two or more process sensors in order from the normal sensor having the highest priority, and sets the optimum target value related to the control using the selected sensor to the local control unit. The optimum operation amount calculation unit outputs an optimum operation amount for the at least one or more actuators when all of the two or more process sensors are abnormal. It is characterized by that.

  According to the present invention, when a plurality of pieces of process sensor information for controlling a certain operation amount are conceivable, a plant-wide optimum control device can be constructed by using a normal sensor among the plurality of sensors. Generally, in feedback control, it is possible to provide robustness by constructing a controller appropriately. Therefore, if there are multiple sensors, the controller can be switched to maintain robustness by feedback control. Robustness against sensor failure can be added.

  The invention according to claim 27 is the plant-wide optimum process control device according to claim 25 or claim 26, wherein at least one of the process sensor abnormality diagnosis unit and the external input sensor abnormality diagnosis unit includes a plurality of sensors. It uses a statistical method such as principal component analysis based on the correlation by.

According to the present invention, a sensor abnormality can be automatically determined by a statistical method, whereby a fully automatic plant-wide robust optimum process control device can be provided.
According to a twenty-eighth aspect of the present invention, in the plant-wide optimum process control apparatus according to the twenty-fifth or twenty-sixth aspect, the process optimization apparatus normally outputs an optimum operation amount as feedforward control, and a process is performed by the process sensor. When the state value deviates from the predetermined fluctuation range, the process sensor abnormality diagnosis unit checks whether the process sensor is abnormal, and if it is not abnormal, switches to feedback control by outputting the optimum target value. It is characterized by that.

  According to the present invention, an optimal amount of operation is given for constraint control in which process output only needs to be within a certain range, rather than control for the purpose and main purpose of following the target value. Therefore, it is possible to provide a plant-wide robust optimum process control device that can add robustness by feedback control only when the constraint is not met.

  The invention according to claim 29 is the plant-wide optimum process control device according to claim 25 or claim 26, wherein hysteresis characteristics are provided in switching between the optimum target value output and the optimum manipulated variable output in the process optimization device. A bidirectional switch is provided.

  According to the present invention, it is possible to prevent the control system from becoming unstable due to chattering in a critical region where judgment of a switching condition such as a sensor abnormality is delicate, and to add robustness to the chattering phenomenon. it can.

  ADVANTAGE OF THE INVENTION According to this invention, control systems, such as the conventional PID control, can be extended seamlessly and a plant-wide optimal control apparatus can be provided.

  By having the conservation law, which is regarded as important in the laws of physics, as a constraint, it is possible to perform plant-wide optimization within the reach of the principle, enabling safe and reliable operation of the process. become.

  The role of each layer is clearly divided by adopting a hierarchical structure in which the conventional control that dynamically compensates the plant is placed in the lower system and the static optimization operation unit of the newly introduced plant is placed in the upper system. Can be made.

  By notifying the human system of a plant abnormality or automatically returning it to the plant, it is possible to perform optimum plant-wide optimal control that is effective not only during normal operation of the plant but also during the abnormality.

  A human system can be appropriately interposed in the automatic control system, and a human-friendly plant-wide optimum control device can be provided.

  It is possible to provide a robust plant-wide process control device that can prevent instability of process control due to process sensor abnormality and can keep the plant in an appropriate state in any case.

<Embodiment 1>
(Configuration of Embodiment 1)
FIG. 1 shows a basic configuration of an embodiment of the present invention when sewage treatment process control is performed in a sewage treatment plant. However, it should be noted that the application target of the embodiment described here is not limited to the sewage treatment process.

  The sewage treatment process shown in FIG. 1 is a sewage advanced treatment process 1 configured as a two-stage circulation nitrification denitrification process. The treated water is first introduced into the settling basin 101, and the first anaerobic tank 102 in this order. Water is discharged through the first aerobic tank 103, the anaerobic tank 104, the second aerobic tank 105, and the final sedimentation basin 106. These tanks 101 to 106 are reaction tanks for treating the water to be treated.

  This advanced sewage treatment process 1 includes, as an actuator, a first sedimentation basin excess sludge extraction pump 111 in the first sedimentation basin 101, and similarly, a step inflow pump 112 between the outlet of the first anaerobic tank 102 and the anoxic tank 104. A blower 113 for supplying oxygen to the first aerobic tank 103, a blower 114 for supplying oxygen to the second aerobic tank 105, a carbon source input pump 115 for supplying a carbon source to the anaerobic tank 104, and a second aerobic tank A circulation pump 116 that circulates non-treated water between 105 and the oxygen-free tank 104, a return sludge pump 117 that returns sludge from the final sedimentation tank 106 to the first sedimentation tank 101, and a flocculant into the second aerobic tank 105. A flocculant charging pump 118 and a final sedimentation tank surplus sludge extraction pump 119 are provided. Each actuator 111-119 operates with a predetermined operation cycle.

  Further, the advanced sewage treatment process 1 is a process sensor for measurement control, an ammonia sensor 121 that measures the ammonia concentration in the first aerobic tank 103, a nitric acid sensor 122 that measures the nitric acid concentration in the anaerobic tank 104, and tanks 102- The MLSS sensor 123 that measures the amount of activated sludge in at least one tank 105 (the oxygen-free tank 104 in the illustrated example), the ammonia sensor 124 that measures the ammonia concentration in the second aerobic tank 105, and the second aerobic tank 105 A phosphoric acid sensor 125 that measures the phosphoric acid concentration, an SS sensor 126 that measures the SS concentration of sludge drawn from the final sedimentation basin 106, a flow rate sensor 127 that measures the amount of excess sludge drawn from the final sedimentation basin 106, and the inflow sewage A plurality of water quality elements such as COD, TN, and TP and a sewage inflow sensor 128 that can measure the amount of water are provided. That. Each of the process sensors 121 to 128 performs measurement at a predetermined measurement cycle. The measurement values measured by the process sensors 121 to 128 are collected and held by the process measurement data collection unit 2 shown in FIG.

  The process measurement data collection unit 2 includes first to seventh data collection units 21 to 27. The first data collection unit 21 collects the nitric acid concentration data of the anoxic tank 104 measured by the nitric acid sensor 122. The second data collection unit 22 collects ammonia concentration data of the first aerobic tank 103 and the second aerobic tank 105 measured by the ammonia sensors 121 and 124. The third data collection unit 23 collects the nitric acid concentration data of the anoxic tank 104 measured by the nitric acid sensor 122. The fourth data collection unit 24 collects the nitric acid concentration data of the anoxic tank 104 measured by the nitric acid sensor 122. The fifth data collection unit 25 collects the activated sludge amount data of the anoxic tank 104 measured by the MLSS sensor 123. The sixth data collection unit 26 collects the phosphate concentration data of the second aerobic tank 105 measured by the phosphate sensor 125. The seventh data collection unit 27 includes the activated sludge amount data of the anoxic tank 104 measured by the MLSS sensor 123, the SS concentration data of sludge extracted from the final sedimentation basin 106, and the flow rate sensor measured by the SS sensor 126. The surplus sludge amount data extracted from the final sedimentation basin 106 measured by 127 is collected. The measurement data collected by the process measurement data collection unit 2 is sent to the local control unit 3.

  The local control unit 3 includes first to seventh local control circuits 31 to 37 corresponding one-to-one to the data collection units 21 to 27 of the process measurement data collection unit 2. The first local control circuit 31 (step inflow pump control circuit) controls the step inflow pump 112. Similarly, the second local control circuit 32 (blower control circuit) controls the blowers 113 and 114, the third local control circuit 33 (carbon source input pump control circuit) controls the carbon source input pump 115, and the fourth local control. The circuit 34 (circulation pump control circuit) is the circulation pump 116, the fifth local control circuit 35 (return sludge pump control circuit) is the return sludge pump 117, and the sixth local control circuit 36 (coagulant charging pump control circuit). Controls the flocculant charging pump 118, and the seventh local control circuit 37 (excess sludge extraction pump control circuit) controls the final sedimentation tank excess sludge extraction pump 119. The local control circuits 31 to 37 receive the process data collected by the data collection units 21 to 27 and the optimum target values supplied from the optimum target value calculation unit 6 and calculate the operation amounts of the actuators 111 to 119. It consists of control logic for making decisions.

  The optimal target value calculation unit 6 collects the optimal evaluation function quantitatively defining and setting the optimality of the process from the optimal evaluation function setting unit 4, and various constraint values from the constraint condition setting unit 5, and collects the eighth data. Data of various water quality and quantity of sewage inflow measured by the sewage inflow sensor 128 is input from the unit 28, and the optimum target value is calculated and transmitted to the local control unit 3. The constraint condition setting unit 5 is a storage amount constraint setting unit 51 that sets a constraint condition related to a storage amount such as a process mass balance or energy balance, and an output that sets a constraint condition such as a water quality or a water amount regulation value that is an output of the process. A constraint setting unit 52 and an operation amount constraint setting unit 53 for setting constraint conditions such as a limit of an operation amount that is a process input and a limit value (Δ limit) of an operation amount change rate are provided. The data collection unit 28 uses inflow sewage data to be substituted into a part of the storage amount constraint equation supplied from the storage amount constraint setting unit 51 when calculating the optimal target value of the optimal target value calculation unit 6. To collect.

(Operation of Embodiment 1)
The operation of the first embodiment will be described with reference to FIG. 1 and FIG. First, the nitric acid data of the oxygen-free tank 104 measured by the nitric acid sensor 122 is stored in the data collection units 21, 23, 24 at a predetermined cycle. Similarly, the ammonia concentration data of the first and second aerobic layers 103 and 105 respectively measured by the ammonia sensors 121 and 124 are stored in the data collection unit 22 at a predetermined cycle. Exactly the same, MLSS data measured by the MLSS sensor 123 is stored in the data collection unit 25, and phosphoric acid data measured by the phosphate sensor 125 is stored in the data collection unit 26. The measured MLSS data of the second aerobic layer 103, the excess sludge SS data measured by the SS sensor 126, and the excess sludge extraction flow rate data measured by the flow rate sensor 127 are stored.

The optimal evaluation function setting unit 4 sets an optimal evaluation function for evaluating the optimality of the sewage advanced treatment process 1. In the advanced sewage treatment process 1, it is required to reduce the operating cost and improve the removal efficiency of nitrogen and phosphorus. However, since these are in a trade-off relationship, an optimal evaluation function is set in consideration of these trade-offs. It is preferable. For example, according to the idea of imposing a levy according to the amount of nitrogen or phosphorus loaded in the effluent water, the effluent water quality cost is converted into monetary amount by converting the improvement of the effluent water amount, and the operation cost is evaluated at the same time. Set an evaluation function J.
J = EC + OC (1)
Here, EC is the discharged water quality cost, OC is the operating cost, and is defined by the following equation, for example.

Here, COD, BOD, SS, TN, and TP represent the chemical oxygen demand, the biochemical oxygen demand, the amount of suspended solids, the total nitrogen concentration, and the total phosphorus concentration in the discharged water, respectively. Are all [kg / m ³ ]. Q _ef [m ³ / d] represents the amount of discharged water. Q _b , Q _ret , Q _circ , Q _ex , Q _pac , and Q _cb are aeration air amount, return amount, circulation amount, surplus sludge extraction amount, flocculant input amount, carbon source input amount, respectively. [M ³ / d]. Q _sludge [kg / d] is the amount of generated sludge. T0 and T indicate a cost evaluation start time and a cost evaluation period, respectively. w _i is a weight corresponding to the cost conversion coefficient, and for example, a value disclosed in Non-Patent Document 6 can be used. Of course, the optimum evaluation function setting unit 4 is not limited to such an evaluation function, and an evaluation function based on various criteria such as user needs or administrative judgment can be used. However, the optimum evaluation function setting unit 4 needs to be able to quantitatively evaluate the criterion of optimality. This concept of the optimum evaluation function is effective as an evaluation function considering such a trade-off because the trade-off of the advanced sewage treatment process 1 can be quantitatively taken into consideration.

  Next, the constraint condition setting unit 5 sets the constraint conditions of the storage amount constraint setting unit 51, the output constraint setting unit 52, and the operation amount constraint setting unit 53. Here, an essential part of the present invention is to have a storage amount constraint setting unit 51. By having the storage amount constraint setting unit 51, the target value calculated by the optimum target value calculation unit 6 described below can be surely reached in principle.

  In the storage amount constraint setting unit 51, description is made with an algebraic equation derived from a differential equation in consideration of mass balance in the processing of the sewage altitude processing process 1.

First, for example, the process model (process simulator) of the advanced sewage treatment process 1 is expressed as follows using the activated sludge models ASM1 to ASM3 described in Non-Patent Document 7, for example.

  Here, x is a vector representing water quality elements such as ammonia, nitric acid, phosphoric acid, COD elements in the reaction tank, SS, MLSS, etc. u is an operation such as aeration amount, circulation amount, return amount, carbon source input amount, etc. D is a vector representing disturbances such as inflow sewage water quality factor and inflow sewage amount, and θ is a vector representing various parameters such as maximum specific growth rate, half-saturation constant, hydrolysis constant, death constant, etc. is there. f is a function representing water quality reaction such as ASM1 to ASM3, physical mixing, supply of oxygen, and the like.

Equation (4) is expressed by connecting a biological reaction tank model and a sedimentation basin model as shown below.
* Bioreaction tank model:

Here, z _* (t) [g / m ³ ] ∈R ⁿ ₊ is a vector representing the water quality element of the reaction tank, n is the number of state variables, and z _{* in} (t) [g / m ³ ] ∈R ⁿ ₊ Is a vector representing the water quality element of the sewage flowing into the target reaction tank, V _* [m ³ ] ∈R ₊ is the volume of the target reaction tank, Q _* (t) [m ³ / day] ∈ R ₊ represents the inflow rate. SεR ^{p × n} is a matrix representing stoichiometry, r (·) εR ^p ₊ is a vector representing the reaction rate, and p is the number of element reaction processes. S _{o2 *} (t) [g / m ³ ] ∈R ₊ is the dissolved oxygen concentration and is one element of z _* (t) (assigned as the second element in ASM1 to ASM3). / S _o2 (t) [g / m ³ ] ∈R ₊ is the amount of saturated dissolved oxygen, KLa [1 / day] ∈R ₊ is the overall oxygen transfer capacity coefficient, Q _B (t) [m ³ / day] ∈R ₊ Represents the amount of aeration, and K [1 / m ³ ] ∈R ₊ represents a constant. Moreover, * represents the character string of each reaction tank, for example, *: = aerobic is an aerobic tank. R ⁿ represents a set of n-th order real numbers, and R ⁿ ₊ represents a set of n-th order positive real numbers.
* Settling pond model:

Here, s _sd (t) [g / m ³ ] ∈R ^S ₊ is a vector representing the soluble water quality element of the settling basin, and s represents the number of soluble water quality elements. x ^b _sd (t) [g / m ³ ] ∈R ^n−S ₊ is a vector representing the floating water quality element of the sedimentation section of the sedimentation basin, and x ^u _sd (t) [g / m ³ ] ∈R ^n-S ₊ is a vector representing the floating water quality element in the supernatant of the sedimentation pond. α is a constant smaller than 1. Q ^w _sdout (t), Q ^r _sdout (t), and Q ^e _sdout (t) represent the excess sludge extraction flow rate, the return flow rate, and the discharge flow rate, respectively.

  By appropriately connecting these biological reaction tank model and sedimentation pond model in accordance with the process configuration of the advanced sewage treatment process 1, equation (4) is obtained.

Now, if the differential value on the left side of equation (4), which is this process model (simulator), is set to zero, a mass balance equation in a steady state is obtained, which is as follows.
f (x, u, θ, d) = 0 (14)

  This expression (14) is an example of a constraint condition expression of the storage amount constraint setting unit 51. The difference between the equation (14) and the sewage treatment process simulator is that the equation (14) is a nonlinear algebraic equation, whereas the process simulator is a differential equation like the equation (4). By employing an algebraic equation instead of a differential equation, the complexity of the optimization calculation of the optimum target value calculation unit 6 described later can be significantly reduced.

  On the other hand, in the process simulator application nonlinear control device disclosed in Patent Documents 1 and 2 and the so-called conventional model predictive control, the differential equation such as the equation (4) is used as a constraint condition, so that the optimization calculation is complicated. It has become. In the present invention, in order to avoid the complexity of the optimization calculation, the local control unit 3 is assigned to compensate for the process dynamics as described above, and only the “steady state” of the process is optimized in the host system. Based on thinking. When the objective is only to optimize the “steady state”, it is not necessary to use a process simulator that can also simulate the “transient state” until reaching the “steady state” as shown in the equation (4). The basic idea of the present invention is to use a “mass balance type” that is easy to handle like the formula.

The equation (14) is a basic equation representing the function of the storage amount restriction setting unit 51. For example, when the activated sludge model ASM2 is adopted, the inflow sewage quality and the inflow sewage amount of the equation (14), etc. Some water quality elements of the disturbance vector d cannot be measured in practice. For example, the activated sludge model ASM2, _X s as a component of _COD, S _a, S f, although in several elements, such as _{S i,} it is usually difficult to measure them directly. Further, as a component of SS and MLSS corresponding to the activated sludge _amount, X _h, X _aut, there are more microorganisms such as _{X pao,} it is difficult to measure even these directly. For this reason, even if a constraint equation regarding the storage amount is given, there may be a situation where the amount of the substance to be stored is unknown. In such a case, for example, a respiration rate test shown in Non-Patent Document 8 can be performed, and these values can be estimated and used. Another method is to estimate the amount of these substances to be stored using a disturbance observer. Not only the sewage treatment process but also a plant-wide optimum process control device having a storage amount constraint setting unit 51 having an external input estimation mechanism that cannot be measured using a concept such as a disturbance observer can be configured. By having such a storage amount constraint setting unit 51, it is possible to avoid a situation in which the process becomes unstable due to control given a water quality target value that can never be achieved, and the process can be stably performed. Can be optimized.

  On the other hand, when trying to estimate the variable value of an external input that should be saved by a disturbance observer, etc., for example, it cannot be estimated sufficiently because the amount or quality of the data has deteriorated, or the disturbance observer configuration conditions are satisfied in the first place. There are cases where it is difficult to estimate these values due to reasons such as lack of reliability, and reliable estimation values cannot be obtained. In such a case, it is conceivable that the expression (14) itself is composed only of measurable variables. Such a concept is well known as a model reduction / simplification for a differential equation such as equation (4). In a differential equation such as equation (4), if the left side is set to 0, an algebraic equation such as equation (14) can be obtained, so the concept of model reduction is also possible for equation (14). It is. For example, in Non-Patent Document 9, from such a viewpoint, the model (process simulator) corresponding to the equation (4) is constructed without finely dividing the COD components. In the model described by the differential equation constructed in this way, by setting the left side = 0, the conserved amount constraint equation can be set only from the measurable variable. This concept can be applied not only to sewage treatment but also to general plants. When a model is created only from such measurable variables, the model is generally considerably simplified, and thus does not become a constraint condition for optimization of the original meaning. However, when only the measurable variable is reliable and the estimated value is not reliable, the target value calculation closer to the optimum can be performed as a result by actively simplifying the structure of the storage amount constraint setting unit 51. .

Next, the output constraint setting unit 52 will be described. In the output constraint setting unit 52, for example, it is possible to set a constraint condition such that a regulation value for water quality that is output or a water quality concentration is never negative. For example, in addition to the equation (14), a constraint condition such as the following equation is set.
0 ≦ x ≦ x _max (15)
Here, x _max is an upper limit value of water quality corresponding to a water quality regulation value or the like.

Next, the operation amount constraint setting unit 53 sets a limit condition that can set a limit value within a range that can be taken by the actuator of the process.
u _min ≦ u ≦ u _max (16)
Here, u _min and u _max correspond to the upper limit value and the lower limit value of the operation amount.

  A combination of the above setting units 51 to 53 is a constraint condition setting unit 5.

Next, the optimum target value calculation unit 6 will be described. The optimum target value calculation unit 6 determines the amount of operation of the actuators (112 to 119) based on the optimum evaluation function set by the optimum evaluation function setting unit 4 and various constraint conditions set by the constraint condition setting unit 5. The optimum target value for the control circuits 31 to 37 is calculated. For this purpose, the following optimization problem may be solved.

  At this time, in the equation (20), the quality of the sewage inflow / the amount of water is required as the amount of the substance to be stored. Therefore, various sewage inflow water quality / water volume data collected and stored in the data collection unit 28 can be averaged over a predetermined period, for example, to give an average inflow condition for a predetermined period. In this way, once the amount of material to be stored is set, this optimization problem can be solved, and the optimum water quality concentration and optimum manipulated variable of the process in the steady state can be determined. However, the optimum operation amount is only in a steady state, and this operation amount is not directly used as the operation amount of the actuator, and the operation amount is determined by the local control unit 3.

As the optimum target value, NO3-N, NH4-N, MLSS, PO4-P, A-SRT and the like may be calculated using the answer to the optimization problem. When written formally,
Optimal Set Points = arg _{N03-N, NH4-N, MLSS, PO4-P, A-SRT} min Cost function subject to (Mass balance, Actuator limit, Output limit)
It becomes.

  The reason why the optimum target value obtained by the optimization calculation is decided instead of determining the optimum operation amount obtained by the optimization calculation is as follows. If the optimum manipulated variable is determined directly, the manipulated variable is not determined so as to have robustness against changes in the process state due to the influence of disturbance or modeling errors in the model used in the optimization calculation. One important purpose of constructing the system is to make the process robust. In order to have robustness, it is necessary to perform feedback control in the local control unit 3, and therefore, here, not the optimum operation amount but the optimum target value for the local control unit 3 is calculated.

  When this optimal target value calculation is actually performed, nonlinear programming can be used. It should be noted that (14) is a nonlinear algebraic equation including a mono-type nonlinearity and a bilinearity in the case of a sewage treatment process. It is necessary to devise so as not to fall into the optimal solution. If this contrivance is performed successfully, a reliable optimum target value can be calculated by using various reliable nonlinear programming algorithms such as the conjugate gradient method.

  On the other hand, when there is no prior a priori information, it may be difficult to set the initial value appropriately. In such a case, a sub-optimal target value can be calculated using metaheuristics such as genetic algorithm (GA). In this case, only suboptimality can be guaranteed, not optimality, but there are advantages such as being difficult to fall into a local minimum solution and being capable of high-speed computation.

Next, the local control unit 3 will be described. The local control unit 3 determines the operation amounts of various actuators. First, the local control circuit 31 takes in the nitric acid data collected and held in the data collecting unit 21 and simultaneously takes in the target value of the optimum nitric acid concentration calculated in the optimum target value calculating unit 6. Then, the optimum step inflow amount can be calculated by, for example, the following PID controller.

Here, K ^st _p , T ^st _I , and T ^st _d are parameters representing the proportional gain, integration time, and derivative time of the PID controller of the step inflow amount, respectively, and s represents both the Laplace operator and the derivative operator. To express. S ^ref _no3 is the target value of the optimum nitric acid concentration calculated by the optimum target value calculation unit 6, and S _no3 (t) is nitric acid concentration data. t is a parameter representing time.

^{_{Here, K st p, T st I}} , it is necessary to appropriately adjust the parameters of ^{T st} _d. For this purpose, for example, gain scheduling can be performed. As a specific method, for example, using a process simulator such as equation (4), a step response test is performed in advance at operating points with a plurality of step inflow amounts, and the nitric acid concentrations corresponding to the operating points are listed. . Further, the parameter value at each operating point is determined from each step response waveform according to a normal adjustment method of the PID controller. An example of this is shown in Table 1.

By _comparing the value of S ^ref _{no3 with} the value of nitric acid concentration in Table 1, the parameters of the PID controller can be switched. Alternatively, the value of the PID controller corresponding to the value of S ^ref _no3 may be calculated by interpolating the parameter values in Table 1. By doing in this way, an appropriate PID parameter is set according to each target value, and as a result, the follow-up characteristic to the target value is improved and the discharged water quality is appropriately maintained even when the inflow sewage load fluctuates. It is possible to improve the disturbance suppression characteristics such as This is an embodiment of the local control circuit 31.

Next, the local control circuit 32 takes in the ammonia data collected and held in the data collection unit 22 and simultaneously takes in the target value of the optimum ammonia concentration calculated in the optimum target value calculation unit 6. Then, the optimal aeration air volume can be calculated by, for example, the following PID controller.

Here, K _b ^p , T _b ^I , and T _b ^d are parameters representing the proportional gain, integration time, and differentiation time of the PID controller of the aeration air volume. S ^ref _nh4 is the target value of the optimum ammonia concentration calculated by the optimum target value calculation unit 6, and S _nh4 (t) is ammonia concentration data.

The parameters K _b ^p , T _b ^I and T _b ^d can be performed by the same method as the parameter adjustment of the PID controller for the step inflow amount.

  As another method, since there is a strong nonlinearity in the relationship between the aeration air volume and the ammonia concentration, local control can be performed using a nonlinear dynamic model. For example, taking the second aerobic tank 105 as an example, the following can be performed.

The relational expression of dissolved oxygen concentration S ₀₂ (t), ammonia concentration S _nh4 and aeration air volume Q _b (t) is almost zero in the biological reaction tank where there is no oxygen flowing into the second aerobic tank 105 and organic matter is already in the previous stage. Assuming that it is consumed, it is given approximately as follows.

Here, μ _A , Y _A , K, K _o2 , and K _nh4 are parameters, Q (t) is a flow rate, V is a reaction tank volume, X _A is a nitrifying bacterium (autotrophic bacterium), and X _tss is a reaction Represents the MLSS concentration in the tank. η is a parameter indicating the percentage of _{X A} present in MLSS. / S _o2 represents the saturated dissolved oxygen concentration. S ⁱⁿ _nh4 (t) represents the concentration of ammonia flowing from the previous reaction tank. Here, it is necessary to determine Q _b (t). However, according to a strict linearization concept, a new input u (t) is introduced, and Q _b (t) is variable-transformed as follows.

同様に、

を（２７）式の入力とみなして、新しい入力ｖ(t)を導入して、以下のように変数変換する。

When this variable conversion is performed, the equation (25) becomes as follows.

Similarly,

Is input as the expression (27), a new input v (t) is introduced, and variable conversion is performed as follows.

When this variable conversion is performed, equation (27) becomes as follows.

  By the above method, the equations (25) and (27) are strictly linearized. Therefore, for example, if u (t) is controlled by PID control, the dissolved oxygen concentration can be controlled. Since v (t) is a link between dissolved oxygen concentration and ammonia concentration in equation (31), this may be designed using, for example, the concept of backstepping.

  The above is the embodiment of the local control circuit 32. By adopting such a method, it is possible to perform local control with robustness that positively considers the nonlinearity of the process.

Next, the local control circuit 33 takes in the nitric acid data collected and held in the data collecting unit 23 and simultaneously takes in the target value of the optimum nitric acid concentration calculated in the optimum target value calculating unit 6. Then, the optimal carbon source input amount can be calculated, for example, by the following PID controller.

Here, K ^cb _p , T ^cb _I , and T ^cb _d are parameters representing the proportional gain, integration time, and differentiation time of the PID controller of the carbon source input amount, respectively. S ^ref _no3 is the target value of the optimum nitric acid concentration calculated by the optimum target value calculation unit 6, and S _no3 (t) is nitric acid concentration data.

Parameters such as K ^b _p , T ^b _I , and T ^b _d can be performed by a method similar to the parameter adjustment of the PID controller for the step inflow amount. It is also possible to configure a local control circuit having robustness such as strict linearization of aeration air amount + backstepping + PID controller.

  The above is the embodiment of the local control circuit 33.

Next, the local control circuit 34 takes in the nitric acid data collected and held in the data collecting unit 24 and simultaneously takes in the target value of the optimum nitric acid concentration calculated in the optimum target value calculating unit 6. Therefore, the optimum circulation amount can be calculated by, for example, the following PID controller.

Here, K ^circ _p , T ^circ _I , and T ^circ _d are parameters representing the proportional gain, integration time, and differentiation time of the PID controller of the circulation amount, respectively.

  From the above description, the step inflow amount of the local control circuit 31, the carbon source input amount of the local control circuit 33, and the circulation amount of the local control circuit 34 are controlled by almost the same algorithm using the same nitric acid concentration data. You can see that it is done. The reason is that the common purpose of “denitrification” is considered as the purpose of controlling these actuators. Of course, these actuators can be controlled for purposes other than "denitrification". However, it is not always a good idea to operate these manipulated variables independently when they have a common purpose as described above. Therefore, for example, it can be carried out as follows.

The denitrification reaction usually proceeds under anoxic conditions, and requires nitric acid and organic matter (measured by COD). The ratio of nitric acid and COD required for this reaction is as follows.

Here, Y _h is a yield of denitrifying bacteria (heterotrophic bacteria).

  In this case, the carbon source input amount for calculating the operation amount by the local control circuit 33 and the step inflow amount for calculating the operation amount by the local control circuit 31 are intended to supply a COD source for denitrification. . The circulation amount calculated by the local control circuit 34 is for the purpose of supplying nitric acid for denitrification. Therefore, in order to keep the operation amount at the ratio of the expression (35), if one of the nitric acid amount and the COD amount is determined, the other is inevitably determined. Therefore, for example, the following control can be performed. First, in the local control circuit 34, the circulation amount is determined by PI control as described above. Next, the required COD amount that maintains the ratio of the equation (35) is calculated. When the calculated COD amount is determined, it is first determined whether or not the calculated required COD amount can be supplied by adjusting the step inflow amount of the local control circuit 31. If the necessary COD amount can be supplied only by adjusting the step inflow amount, the necessary COD amount is supplied by the step inflow, and the carbon source is not input to the local control circuit 33. If it is determined that the required COD amount cannot be supplied only by adjusting the step inflow amount, the shortage is calculated, and the shortage is supplied by turning on the carbon source of the local control circuit 33. By adopting such an algorithm, optimal denitrification control without excess or deficiency can be performed. This is an embodiment of the local control circuit 31, the local control circuit 33, and the local control circuit 34.

Next, the local control circuit 35 takes in the MLSS data collected and held in the data collecting unit 25 and simultaneously takes in the target value of the optimum MLSS concentration calculated in the optimum target value calculating unit 6. Then, the optimal return amount can be calculated by, for example, the following PID controller.

Here, K ^ret _p , T ^ret _I , and T ^ret _d are parameters representing the proportional gain, integration time, and derivative time of the PID controller of the return amount. X ^ref _tss is the target value of the optimal MLSS concentration calculated by the optimal target value calculation unit 6, and X _tss (t) is the MLSS concentration data. Parameters such as K ^ret _p , T ^ret _I , and T ^ret _d can be performed by the same method as the parameter adjustment of the PID controller for the step inflow amount. As a result, the optimum return sludge amount can be adjusted. The above is the embodiment of the local control circuit 35.

Next, the local control circuit 36 takes in the phosphoric acid data collected and held in the data collecting unit 26 and, at the same time, the optimum phosphoric acid concentration calculated in the optimum target value calculating unit 6 as described above. Capture target value. Then, the optimal flocculant input amount can be calculated by, for example, the following PID controller.

_{^{Here, K po4 p, T po4 I}} , T po4 d each proportional gain of the PID Kontora flocculant dosages, integration time, and a parameter representing the derivative time. Further, S ^ref _po4 is a target value of the optimum phosphoric acid concentration calculated by the optimum target value calculating unit 6, and S _po4 (t) is phosphoric acid concentration data.

K ^{_{po4 p,} T po4} I, parameters such as ^{T PO4} _d can be performed by the parameter adjustment and similar methods of the PID controller in step inflow. This flocculant charging control can also be performed using a dynamic model as described in the explanation of the local control of the aeration air volume.

  Thereby, optimal flocculant injection | throwing-in can be performed, and chemical cost can be reduced while performing appropriate phosphorus removal. It should be noted here that if the phosphoric acid concentration measured is lower than the target value of the phosphoric acid concentration set by the optimum target value calculation unit 6, the flocculant is not added as a result. That is. Therefore, the local control circuit having such an optimal target value setting value is an intelligent controller that operates only when it is determined that the medicine needs to be charged.

Next, when the local control circuit 37 takes in the MLSS data in the aeration tank collected and held in the data collecting unit 27, the SS data in the extracted sludge, and the extracted sludge amount data, as described above. At the same time, the target value of the optimum SRT (solids residence time) or A-SRT (solids residence time under aerobic conditions) calculated by the optimum target value calculation unit 6 is fetched. Then, the optimal excess sludge extraction amount can be calculated by, for example, the following equation SRT control.

Here, X ^reactor _tss (t), V, X ^waste _tss (t), and SRT _ref are the MLSS concentration in the biological reaction tank, the capacity of the biological reaction tank, the SS concentration of excess sludge, and the optimum target value calculation unit 6, respectively. Is the target value of the SRT supplied by.

  The optimum amount of excess sludge extraction can be calculated by equation (38). In equation (38), excess sludge is continuously extracted. However, in actual sewage treatment plants, excess sludge extraction is often controlled by a timer, so that it is easily extracted. Many. One reason for this is that when sludge is continuously extracted, the sedimentation of the sludge may be poor, but with the timer control, when the sludge is sufficiently settled, the sludge is extracted all at once, so the operation of extracting excess sludge is successful. Because it works. Therefore, for example, by newly installing a sludge interface meter, it is possible to judge the sludge settling condition and to extract sludge intermittently. For example, the upper limit threshold level of the sludge interface is set, and the excess sludge extraction amount of the equation (38) is integrated until the interface exceeds the upper limit, and the amount of excess sludge integrated when the upper limit is exceeded. It is possible to perform a control such that a certain amount of time is pulled out. By doing in this way, the excess sludge extraction control suitable for an actual process can be implemented. Further, since the optimum SRT value is supplied from the optimum target value calculation unit 6, the optimum amount of sludge can be controlled. The above is the embodiment of the local control circuit 37.

  According to the series of operations as described above, optimal plant-wide sewage treatment process control can be performed. However, the optimum target value in the series of operations described above should actually differ depending on the quantity and quality of the incoming sewage. In particular, the amount and quality of influent sewage usually changes due to differences in weather such as rainy weather and clear weather, differences between holidays and weekdays, differences between night and daytime, and differences in season. Therefore, it is more realistic to repeat the optimal target value calculation at an appropriate cycle, and the optimal target value calculation can be performed at a predetermined constant cycle. In addition, when the above-mentioned weather, day of the week, time zone, and season are different, the optimum target value calculation can be recalculated instead of a fixed period. By such optimal target value recalculation, it is possible to optimally control the sewage advanced treatment process in a plant-wide manner according to the inflow load situation of the sewage.

(Effect of Embodiment 1)
In the above embodiment, by installing an optimum target value calculation device having a substance amount storage constraint condition as a supervisor in a host system of a local control unit such as conventional PID control, the following effects can be obtained. Can do.

(1) It is possible to provide an optimum sewage treatment control device that optimizes the quality of effluent water and the operation cost.
(2) It is possible to avoid the danger of giving a target value that cannot be reached in principle due to the substance amount storage constraint condition.
(3) It is possible to provide a seamless plant-wide sewage treatment process control device by using a conventional control device of a local control unit as it is.
(4) The local control device and the optimum target value calculation device can be set independently. For example, the local control device takes into consideration the robustness and stability of the plant, and the optimum target value calculation device is used for plant-wide optimization of the plant. It is possible to clarify the division of roles such as

<Embodiment 2>
(Configuration of Embodiment 2)
3 and 4 show the second embodiment. FIG. 3 shows a sewage altitude treatment process 1 similar to that shown in FIG. 1 exemplified as a control target, and the same or similar components as those in the sewage altitude treatment process 1 of FIG. The difference between the two is that, in the process of FIG. 3, the second aerobic tank 105 is additionally provided with an alkaline agent charging pump 120 as one of the actuators. 105 is additionally provided with an alkalinity / pH sensor 129 for measuring alkalinity or pH, and an ORP sensor 130 for measuring an ORP value. The other is no different from the process of FIG.

  FIG. 4 is a block diagram showing a plant-wide optimum process control apparatus that controls the advanced sewage treatment process 1 through the actuators 111 to 120 described in FIG. Here, the process measurement data collection unit 2 is shown as one block including the data collection units 21 to 27 and the data collection unit 28 shown in FIG. The same applies to the local control unit 3.

  4 includes a process measurement data collection unit 2, a local control unit 3, an optimum evaluation function setting unit 4, a constraint condition setting unit 5, an optimum target value calculation unit 6, a process abnormality determination unit 7, and a process. An abnormality notifying unit 8 and a process abnormality diagnosing unit 9 are included. The individual functions of the process measurement data collection unit 2, the local control unit 3, the optimum evaluation function setting unit 4, the constraint condition setting unit 5, and the optimum target value calculation unit 6 are as described above.

  The process abnormality determination unit 7 inputs various measurement data collected by the process data collection unit 2 and the optimum target value calculated by the optimum target value calculation unit 6, and determines a process abnormality. The process abnormality notification unit 8 notifies the process operator and manager of the process abnormality / normal state determined by the process abnormality determination unit 7. When the process abnormality determination unit 7 determines that an abnormality has occurred, the process abnormality diagnosis unit 9 investigates the cause of the abnormality and diagnoses the process abnormality. The process abnormality control unit 10 performs new control so as to restore the cause of the process abnormality diagnosed by the process abnormality diagnosis unit 9.

(Operation of Embodiment 2)
The operation of the second embodiment will be described with reference to FIGS. The normal operation of the present embodiment is almost the same as that of the first embodiment except for the part related to process abnormality diagnosis and abnormality determination. Therefore, only the operation of the part related to the process abnormality will be described below.

First, the process measurement data collection unit 2 collects and holds data of the alkalinity / pH sensor 129 and the ORP sensor 130 at a predetermined period. The process abnormality determination unit 7 inputs the optimum target value calculated by the optimum target value calculation unit 6 and the measurement data held in the process measurement data collection unit 2. Then, the error between the optimum target value of each sensor and the measurement data is integrated as in the following equation.

Here, X _ref and X (t) are a target value and a measured value of a certain water quality to be controlled.

In the second embodiment, since the constraint condition setting unit 5 considers mass balance, if the process is operating normally, the error on the right side of the equation (39) is 0 (zero) as time elapses. Should converge to. In other words, the target value cannot be reached. Therefore, the value on the left side does not increase indefinitely. Therefore, a certain threshold level K is set, and normality / abnormality of the process is determined by the following equation.
if | IE | ≦ K then normal
if | IE |> K then abnormal (40)

Based on such determination, it can be determined whether the process is normal or abnormal. However, in a pure integrator such as the equation (39), the left side may increase infinitely even if a slight error remains, so an extremely small positive forgetting factor α is introduced, You may change like following Formula.

  By correcting the determination criteria in this way, it is less likely to be influenced by “slight deviation from the target value” that can often occur in practice, so that the normality / abnormality of the process can be determined more accurately. This is an embodiment of the process abnormality determination unit 7.

  Based on the determination result of the process abnormality determination unit 7, the process abnormality notification unit 8 notifies the operator or manager of the sewage altitude treatment process 1 whether the process is normal or abnormal. This is an embodiment of the process abnormality notification unit 8 and an embodiment of a plant-wide optimum sewage treatment process having a process abnormality notification unit. By notifying the operator of the normality / abnormality of such a process, the process can be operated more safely and with higher reliability.

On the other hand, if not only the normality / abnormality of the process can be notified but also the cause of the process can be listed, the plant operation with higher reliability and safety becomes possible. Therefore, the process abnormality diagnosis unit 9 is useful. The process abnormality diagnosis unit 9 is performed in response to the determination result of the process abnormality determination unit 7. The process abnormality diagnosis unit 9 operates when the process abnormality determination unit 7 determines that the process is abnormal. The process abnormality diagnosis unit 9 has a list of process abnormality factors. Here, the case where the local control unit 3 controls the ammonia concentration in the second aerobic tank 105 will be described with reference to Table 2.

As shown in Table 2, for example, as a factor when ammonia concentration control by the amount of aeration is not successful,
(1) Abnormal flow rate due to a malfunction of the aeration device,
(2) The abnormality of the ammonia sensor itself,
(3) Nitrification inhibition such as when the pH is extremely acidic,
And so on. Therefore, first, these possible factors are listed in the vertical direction of the table. Next, the minimum and maximum values of the normal range of these values are written into the table. The maximum value and the minimum value may be given as fixed values in advance, or variable values may be entered according to the process status. For example, when the optimum target value of the ammonia concentration calculated by the optimum target value calculation unit 6 is 1 [mg / L] and the value of the ORP sensor 130 is on the sufficient oxidation side, for example, 130 [mV] or more, ammonia Since the concentration cannot be extremely high, the minimum value and the maximum value are periodically written in a table in accordance with a rule such that the maximum value of the ammonia concentration in this case is 12 [mg / L]. At the same time, the actual value currently measured is written into the table. Furthermore, when the measured actual value is between the maximum and minimum values, “normal” is written, and when the measured value is outside the maximum and minimum values, “abnormal” is written. Write. For example, in Table 2, since the pH is extremely acidic, “abnormal” is written in this item. This is the operation of the process abnormality diagnosis unit 9. Further, the process abnormality notification unit 8 notifies that the process is abnormal and that the cause is highly likely that the pH is extremely acidic. If the item “abnormal” does not appear in Table 2, the fact that the process abnormality factor is not the item in Table 2 is notified, and at the same time, notification of the abnormality factor unknown is performed.

  As described above, not only the operator and the manager can be notified that the process is abnormal, but also the factor can be notified, and the process can be operated more safely and with high reliability.

  On the other hand, if the process abnormality cause can be found by such process abnormality diagnosis, the countermeasure can be implemented at the same time. By providing the process abnormality control unit 10, this can be achieved.

  The process abnormality control unit 10 performs control such that the process returns to normal based on the cause diagnosed by the process abnormality diagnosis unit 9. As shown in Table 2, if the cause of the process abnormality is nitrification inhibition due to the process being too acidic, the process may be neutralized. Therefore, the amount of alkali agent charged by the alkali agent charging pump 120 is controlled. First, from Table 2, an intermediate value between the maximum value and the minimum value of pH is set as a target value. The control of the amount of the alkaline agent input is performed using the same PID control as has been repeatedly described so far. In this case, since the pH sensor 129 is provided as a sensor, PID control may be performed on the output of the pH sensor. When the pH returns to a predetermined value, the alkaline agent charging PID control is stopped and the normal operation is resumed.

  If the cause diagnosed by the process abnormality diagnosis unit 9 is abnormality of the ammonia sensors 121 and 124, the process abnormality control unit 10 switches to the aeration air amount PID control using the ORP sensor 130, Announcement is made to prompt the process abnormality notification unit 8 to repair the ammonia sensor. If the cause diagnosed by the process abnormality diagnosis unit 9 is an abnormality in the amount of aeration, the air diffuser for aeration is automatically cleaned.

  The above is the operation of the process abnormality control unit 10. This process abnormality control unit 10 performs control at the time of abnormality, and at the same time, the process abnormality notification unit 8 prompts the removal of the abnormality factor, and when the operator or the administrator eliminates the process abnormality factor, this process abnormality control A promise is made to send some signal to the unit 10, and when this signal is received, the control of the process returns to the normal control.

  The above is the embodiment of the runt-wide optimum process control device incorporating the process abnormality control device. Thereby, even when the process is abnormal, the process can be automatically returned to the normal state, and the process operation with higher reliability and safety can be performed.

(Effect of Embodiment 2)
The main effects obtained from the present embodiment include the following effects in addition to the effects of the first embodiment. That is, not only when the advanced sewage treatment process is operating normally, but also when the process is in an abnormal state due to some factor, the process can be optimally controlled on a plant-wide basis. Therefore, the process can be operated almost completely automatically.

<Embodiment 3>
(Configuration of Embodiment 3)
FIG. 5 is an explanatory diagram of the third embodiment. The configuration of the third embodiment is almost the same as either of the processes shown in FIG. 1 or FIG. The difference is that a manual target value setting unit 11 is provided as shown in FIG.

(Operation of Embodiment 3)
The operation of the third embodiment will be described with reference to FIG. As described in the section of operation of the first embodiment, the optimal target value calculation unit 6 calculates the optimal target value of each water quality sensor, but this value may not be satisfactory for the operator. For example, in the optimum evaluation function setting unit 4 shown in FIG. 2, an evaluation function such as Expression (18) or Expression (19) is set, but the operator does not consider the situation that is not considered in these expressions. May be evaluated empirically. In such a case, there may be a case where the optimum target value obtained by the optimum target value calculation unit 6 is desired to be corrected. Therefore, the optimum target value obtained by the optimum target value calculation unit 6 is given to the local control unit 3 and simultaneously presented to the operator 13. When the operator 13 wants to correct this target value, the optimum target value calculation unit 6 is switched to the manual target value setting unit 11. Then, the manual target value setting unit 11 sets a target value that the operator thinks desirable. Next, the manual target value determination unit 12 determines whether the target value is consistent with the conditions of the storage amount constraint setting unit 51 of FIG. If the set target values are contradictory, an announcement is made to the operator 13 to change the target values. If there is no contradiction, the target value is transmitted to the local control unit 3. The above is the target value setting mode according to the third embodiment.

(Effect of Embodiment 3)
In addition to the effect of the first embodiment, the effect obtained from the present embodiment is that the human system and the automatic control system can be coordinated in the operation of the sewage treatment plant. When it is desired to intervene a human system, partial automatic control with assistance by automatic control can be performed.

<Embodiment 4>
(Configuration of Embodiment 4)
FIG. 6 is an explanatory diagram of the fourth embodiment. The configuration of the fourth embodiment is almost the same as either of the processes shown in FIG. 1 or FIG. The only difference is that the local control unit 3 is disconnected as shown in FIG.

(Operation of Embodiment 4)
The operation of the fourth embodiment will be described with reference to FIG. As described in the section of the operation of the first embodiment, the optimum target value calculated by the optimum target value calculation unit 6 is sent to the local control unit 3 for automatic control, but the operator of the sewage treatment process And it is not always the case that automatic control is best for the administrator. An operator who is proficient in process operation may wish to manually control the various actuators 111 to 120. Even in such a case, it is very laborious for the operator to consider all the values of the water quality that should be set. Therefore, the optimum target value calculated by the optimum target value calculation unit 6 is presented to the operator 13, and at the same time, the data collected and stored by the process measurement data collection unit 2 is presented to the operator 13 as time series data. The operator 13 manually determines the operation amounts of the various actuators 111 to 120 by simultaneously viewing the optimum target value and the actual measurement data on the display. In this case, the operation amount of only a part of the actuators 111 to 120 may be determined manually. Since the operation amount time series data and the water quality time series data are always presented to the operator 13 together with the target value, this is a visual feedback system through humans. Can be substituted. The above is the embodiment of the local control by the human system.

(Effect of Embodiment 4)
In the fourth embodiment, in the same way as the third embodiment, in addition to the effects of the first embodiment, the human system and the automatic control system can be coordinated in the operation of the sewage treatment plant. If you want to perform fully automatic control and want to intervene human system, you can perform partial automatic control with support by automatic control.

<Embodiment 5>
(Configuration of Embodiment 5)
FIG. 7 is a configuration diagram of the fifth embodiment. The fifth embodiment has robustness against sensor abnormality, and provides a plant-wide optimum process control apparatus that can continue control while avoiding failure of process control even if sensor abnormality occurs for some reason. Is. 8 is an explanatory diagram showing a detailed configuration of the sewage altitude treatment process 1 in FIG. 7, and FIG. 9 is an explanatory diagram showing a detailed configuration of the process measurement data collection unit 2 and the local control unit 3 in FIG. In these drawings, the same components as those already described in FIGS. 1 to 3 are denoted by the same reference numerals.

  In FIG. 7, measurement data from the sewage altitude treatment process 1 is collected by the process measurement data collection unit 2, and the process measurement data collection unit 2 sends the collected data to the local control unit 3 and the process sensor abnormality diagnosis unit 14. It has become. Further, specific external input measurement data from the sewage altitude treatment process 1 is collected by the external input measurement data collection unit 15, and the external input measurement data collection unit 15 sends this to the external input sensor abnormality diagnosis unit 16. Yes.

  The external input sensor abnormality diagnosing unit 16 and the process sensor abnormality diagnosing unit 14 perform sensor abnormality diagnosis based on the input measurement data, and switch contacts of the switch units SW1 and SW2, respectively.

  The constraint condition setting unit 5 inputs data from the external input measurement data collecting unit 15 when the sensor is normal via the switch unit SW1, and inputs data from the manual analysis data storage unit 17 when the sensor is abnormal. ing. The manual analysis data storage unit 17 stores manual analysis data on the sewage inflow water quality obtained in advance.

  The apparatus according to the present embodiment includes a process optimization unit 18 that performs process optimization when the sensor is normal and when the sensor is abnormal based on the settings of the optimum evaluation function setting unit 4 and the constraint condition setting unit 5. The process optimizing unit 18 outputs an optimal target value when the sensor is normal to the local control unit 3 via the switch unit SW2, and an optimal operation amount when the sensor is abnormal via the switch unit SW2. And an optimum manipulated variable calculator 19 that directly outputs to the actuators 112 to 119 of the advanced sewage treatment process 1.

  The sewage altitude treatment process 1 shown in FIG. 8 is substantially the same as that shown in FIG. 1 except that the measurement data of the sewage inflow sensor 128 is output to the external input measurement data collection unit 15. It is a point.

  As shown in FIG. 9, the process measurement data collection unit 2 includes a first data collection unit 21 that collects data from the nitric acid sensor 122 in the anoxic tank 104, an ammonia sensor 121 in the aerobic tank 103, and a favorable condition. A second data collection unit 22 that collects data from the ammonia sensor 124 in the air tank 105, a third data collection unit 23 that collects data from the nitric acid sensor 122 in the oxygen-free tank 104, and an oxygen-free tank In the aerobic tank 105, the 4th data collection part 24 which collects the data from the nitric acid sensor 122 in 104, the 5th data collection part 25 which collects the data from the MLSS sensor 123 in the anoxic tank 104, and A sixth data collection unit 26 that collects data from the phosphoric acid sensor 125, data from the MLSS sensor 123 in the anoxic tank 104, and SS data of the final sedimentation site 106. It has data from service 126, and a seventh data collecting section 27 for collecting data from the flow sensor 127 provided in the final sedimentation areas 106 bottom side.

  The local control unit 3 includes a first local control circuit 31 that controls the step inflow pump 112 as a step inflow pump control circuit, a second local control circuit 32 that controls the blowers 113 and 114 as a blower control circuit, A third local control circuit 33 that controls the carbon source input pump 115 as a carbon source input pump control circuit, a fourth local control circuit 34 that controls the circulation pump 116 as a circulation pump control circuit, and a return sludge pump control circuit A fifth local control circuit 35 for controlling the return sludge pump 117, a sixth local control circuit 36 for controlling the flocculant charging pump 118 as a flocculant charging pump control circuit, and an excess sludge as a surplus sludge stripping pump control circuit And a seventh local control circuit 37 for controlling the drawing pump 119. .

  And each of these local control circuits 31-37 inputs the measured value of the data collection parts 21-27 and the optimal target value from the optimal target value calculating part 6, and calculates the operation amount calculated based on these inputs. It outputs to the actuators 112-119.

  Each measurement value of the data collection units 21 to 27 is also output to the process sensor abnormality diagnosis unit 14. The process sensor abnormality diagnosis unit 14 performs abnormality diagnosis of the process sensor based on these inputs and switches the contact of the switch unit SW2.

(Operation of Embodiment 5)
Next, the operation of the fifth embodiment will be described with reference to FIGS. The method of setting the optimum evaluation function setting unit 4 and the constraint condition setting unit 5 is the same as that described in the first embodiment, but will be described again for convenience of explanation.

The optimal evaluation function setting unit 4 sets an optimal evaluation function J for evaluating the optimality of the sewage advanced treatment process 1 as shown in, for example, equation (1).
J = EC + OC (1)

Here, EC is the discharged water quality cost, OC is the operating cost, and is defined by the following equation, for example.

Here, COD, BOD, SS, TN, and TP represent the chemical oxygen demand, the biochemical oxygen demand, the amount of suspended solids, the total nitrogen concentration, and the total phosphorus concentration in the discharged water, respectively. Are all [kg / m ³ ]. Q _ef [m ³ / d] represents the amount of discharged water. Q _b , Q _ret , Q _circ , Q _ex , Q _pac , and Q _cb are aeration air amount, return amount, circulation amount, surplus sludge extraction amount, flocculant input amount, carbon source input amount, respectively. [M ³ / d]. Q _sludge [kg / d] is the amount of generated sludge. T0 and T indicate a cost evaluation start time and a cost evaluation period, respectively. w _i is a weight corresponding to the cost conversion coefficient, and for example, a value disclosed in Non-Patent Document 6 can be used.

  Next, the constraint condition setting unit 5 sets the constraint conditions of the storage amount constraint setting unit 51, the output constraint setting unit 52, and the operation amount constraint setting unit 53.

  In the storage amount constraint setting unit 51, description is made with an algebraic equation derived from a differential equation in consideration of mass balance in the processing of the sewage altitude processing process 1.

First, for example, the process model (process simulator) of the advanced sewage treatment process 1 is expressed as follows using the activated sludge models ASM1 to ASM3 described in Non-Patent Document 7, for example.

  Here, x is a vector representing water quality elements such as ammonia, nitric acid, phosphoric acid, COD elements in the reaction tank, SS, MLSS, etc. u is an operation such as aeration amount, circulation amount, return amount, carbon source input amount, etc. D is a vector representing disturbances such as inflow sewage water quality factor and inflow sewage amount, and θ is a vector representing various parameters such as maximum specific growth rate, half-saturation constant, hydrolysis constant, death constant, etc. is there. f is a function representing water quality reaction such as ASM1 to ASM3, physical mixing, supply of oxygen, and the like.

Equation (4) is expressed by connecting a biological reaction tank model and a sedimentation basin model as shown below.
* Bioreaction tank model:

Here, z _* (t) [g / m ³ ] ∈R ⁿ ₊ is a vector representing the water quality element of the reaction tank, n is the number of state variables, and z _{* in} (t) [g / m ³ ] ∈R ⁿ ₊ Is a vector representing the water quality element of the sewage flowing into the target reaction tank, V _* [m ³ ] ∈R ₊ is the volume of the target reaction tank, Q _* (t) [m ³ / day] ∈ R ₊ represents the inflow rate. SεR ^{p × n} is a matrix representing stoichiometry, r (·) εR ^p ₊ is a vector representing the reaction rate, and p is the number of element reaction processes. S _{o2 *} (t) [g / m ³ ] ∈R ₊ is the dissolved oxygen concentration and is one element of z _* (t) (assigned as the second element in ASM1 to ASM3). / S _o2 (t) [g / m ³ ] ∈R ₊ is the amount of saturated dissolved oxygen, KLa [1 / day] ∈R ₊ is the overall oxygen transfer capacity coefficient, Q _B (t) [m ³ / day] ∈R ₊ Represents the amount of aeration, and K [1 / m ³ ] ∈R ₊ represents a constant. Moreover, * represents the character string of each reaction tank, for example, *: = aerobic is an aerobic tank. R ⁿ represents a set of n-th order real numbers, and R ⁿ ₊ represents a set of n-th order positive real numbers.
* Settling pond model:

Here, s _sd (t) [g / m ³ ] ∈R ^S ₊ is a vector representing the soluble water quality element of the settling basin, and s represents the number of soluble water quality elements. x ^b _sd (t) [g / m ³ ] ∈R ^n−S ₊ is a vector representing the floating water quality element of the sedimentation section of the sedimentation basin, and x ^u _sd (t) [g / m ³ ] ∈R ^n-S ₊ is a vector representing the floating water quality element in the supernatant of the sedimentation pond. α is a constant smaller than 1. Q ^w _sdout (t), Q ^r _sdout (t), and Q ^e _sdout (t) represent the excess sludge extraction flow rate, the return flow rate, and the discharge flow rate, respectively.

  By appropriately connecting these biological reaction tank model and sedimentation pond model in accordance with the process configuration of the advanced sewage treatment process 1, equation (4) is obtained.

Now, if the differential value on the left side of equation (4), which is this process model (simulator), is set to zero, a mass balance equation in a steady state is obtained, which is as follows.
f (x, u, θ, d) = 0 (14)

  This expression (14) is an example of a constraint condition expression of the storage amount constraint setting unit 51. The difference between the equation (14) and the sewage treatment process simulator is that the equation (14) is a nonlinear algebraic equation, whereas the process simulator is a differential equation like the equation (4). This is as described in the first embodiment. It should be noted that the mass balance equation (14) includes not only the various water qualities of the sewage treatment process but also the operation amount. This fact will be used later.

Next, the output constraint setting unit 52 will be described. In the output constraint setting unit 52, for example, it is possible to set a constraint condition such that a regulation value for water quality that is output or a water quality concentration is never negative. For example, in addition to the equation (14), a constraint condition such as the following equation is set.
0 ≦ x ≦ x _max (15)

Here, x _max is an upper limit value of water quality corresponding to a water quality regulation value or the like.

Next, the operation amount constraint setting unit 53 sets a limit condition that can set a limit value within a range that can be taken by the actuator of the process.
u _min ≦ u ≦ u _max (16)

Here, u _min and u _max correspond to the upper limit value and the lower limit value of the operation amount. A combination of the above setting units 51 to 53 is a constraint condition setting unit 5.

  After the above setting is performed, process measurement data collection from the sewage altitude treatment process 1 by the process measurement data collection unit 2 is performed in a predetermined cycle, and control by the local control unit 3 is performed.

  Next, the external input sensor abnormality diagnosis unit 16 performs abnormality diagnosis of the sensor value using the time series data of the inflow sewage quality and the inflow sewage amount collected by the external input measurement data collection unit 15. Various methods are conceivable as specific methods, for example, as follows. As an example, the ammonia concentration in the influent sewage is taken up.

  First, the typical value of ammonia concentration in the influent sewage is investigated. Normally, the value of ammonia concentration is about 15 [mg / L] to 40 [mg / L] for ordinary municipal sewage. Therefore, an upper limit value and a lower limit value that the ammonia concentration can take are set. For example, the lower limit is 10 [mg / L], and the upper limit is 40 [mg / L].

In addition, the maximum width of the change in ammonia concentration is set in advance. For example, assuming that collection is performed at a cycle of 10 minutes, the maximum change rate of the ammonia concentration in 10 minutes is set to 10 [mg / L]. Furthermore, the autocorrelation function of the time series data of ammonia concentration is calculated, and it is examined how much past data and current data are correlated. Then, for example, a time lag in which the correlation coefficient is 0.8 (when 1 is set in the case of complete correlation) is observed. For example, it is assumed that data up to 30 minutes before is usually obtained when the correlation coefficient is 0.8 or more. Based on these prior investigations, a map for judging the normality / abnormality of the ammonia sensor is prepared as shown in Table 3, for example.

  In Table 3, when all items are within the normal range, it is determined that “the sensor is normal”. Conversely, if all items are in the abnormal range, it is determined that “the sensor is abnormal”. If any item is in the abnormal range, create a mechanism to announce sensor check. For example, the external input sensor abnormality diagnosis unit 16 incorporating such logic is used to diagnose normality / abnormality of measured values such as inflow sewage quality and inflow sewage amount.

  If the external input sensor abnormality diagnosing unit 16 determines that the sensor is normal, the external input sensor abnormality diagnosing unit 16 sets the contact point of the switch unit SW1 to the a side, and the inflow sewage amount collected by the external input measurement data collecting unit 15 And processing data on the time series data of the inflow sewage quality is input to the item d of the storage amount constraint setting unit 51.

  On the other hand, if it is determined that the sensor is abnormal, the external input sensor abnormality diagnosis unit 16 sets the contact of the switch unit SW1 to the b side, and sets the data stored in the manual analysis data storage unit 17 to d of the storage amount constraint setting unit 51. Appropriate conversion is performed so as to have a form corresponding to, and input to the term of d is performed. In the manual analysis data storage unit 17, for example, data such as COD, ammonia nitrogen, and phosphoric acid phosphorus analyzed once a day are stored in advance.

  If it is not possible to determine whether the sensor is normal or abnormal, the operator or administrator determines whether the sensor is normal or abnormal according to the announcement, and either manually operates the switch unit SW1. Select one input.

  The process sensor abnormality diagnosis unit 14 performs abnormality diagnosis of the measurement value by the process sensor based on the time series data of the process value collected by the process measurement data collection unit 2. This abnormality diagnosis may be performed by a method similar to that of the external input sensor abnormality diagnosis unit 16 described above, but by a method using a statistical method such as principal component analysis based on correlation by a plurality of sensors as follows. It can be carried out.

  That is, some of the process sensors often have a very strong correlation. For example, an ammonia sensor, a DO sensor, and an ORP sensor are often correlated with each other, and the correlation is particularly strong under aerobic conditions. Therefore, principal component analysis is performed on these three time series data or processed time series data. Then, when the sensor is normal, the value of the first principal component is, for example, about 0.9 (the sum of the first to third principal components is about 1). Therefore, for example, 0.75 is set as the threshold level, and if the value of the first principal component becomes less than 0.75, it is determined that one of the sensors is abnormal. For example, if the sensor is normal, a flag such as FB (feedback) is set, and if the sensor is abnormal, a flag such as FF (feed forward) is set to hold the sensor state.

  The process optimization unit 18 calculates an optimum target value and an optimum operation amount of the process based on the optimum evaluation function set by the optimum evaluation function setting unit 4 and various constraint conditions set by the constraint condition setting unit 5. . Here, it should be noted that the optimization problem is the same as that given in the first embodiment, but in the first embodiment, the local control unit 3 is selected from the solutions of the optimization problem. In this fifth embodiment, not only the optimum target value but also the optimum operation amount is calculated.

In the first embodiment, it is asserted that it is more robust against plant variation to obtain an optimum target value and perform feedback control in the local control unit 3 than to obtain an optimum manipulated variable and perform feedforward control. However, this is based on the premise that “the sensor is normal”. If the sensor is abnormal and the measurement information of the sensor cannot be trusted, feedback control using this will adversely affect the process. Therefore, the fifth embodiment proposes switching between feedback control and feedforward control according to the normal / abnormal state of the sensor. The optimization problem to be solved here is the first embodiment.
And is given below.

The optimization problem can be solved if the constraint condition setting unit 5 and the optimal evaluation function setting unit 4 are respectively provided with the constraint condition and the evaluation function, and the optimal water concentration and the optimum operation amount of the process in the steady state. Is obtained. For the optimal target value, NO3-N, NH4-N, MLSS, PO4-P, A-SRT, etc. can be calculated using the answer to this optimization problem. When written formally,
Optimal Reference Set Points = arg _{N03-N, NH4-N, MLSS, PO4-P, A-SRT} min Cost function subject to (Mass balance, Actuator limit, Output limit)
It becomes.

The optimal amount of operation is the answer to this optimization problem itself.
Optimal Actuator Set Points = arg _{Qstep, Qb, Qpac, Qcb, Qcirc, Qret, Qwaste} min Cost function subject to (Mass balance, Actuator limit, Output Limit)
It becomes.

  That is, the problem to be solved is a common optimization problem, but what is adopted in the solution is different. When Optimal Reference Set Points is selected, the result is feedback + feedforward control, and when Optimal Actuator Set Points is selected, pure feedforward control is performed.

  In general, feedback control is said to have the effect of increasing robustness, but this is robust to (1) process model uncertainty and (2) process contamination, and if the sensor becomes abnormal. It is not that robust. Therefore, when the process sensor abnormality diagnosis unit 14 determines that the process sensor state is FF and is abnormal, the process sensor abnormality diagnosis unit 14 turns ON the contact d side of the switch unit SW2 to select Optimal Actuator Set Points (at this time, the contact c) The side is OFF). Thereby, for the actuators 112 to 119, that is, the step inflow pump 112, the blowers 113 and 114, the carbon source input pump 115, the circulation pump 116, the return sludge pump 117, the flocculant input pump 118, and the excess sludge extraction pump 119. Each optimum operation amount calculated by the optimum operation amount calculation unit 19 is directly given.

  Here, usually, in the sewage treatment process, there is often a constant ratio control that keeps the ratio of the manipulated variable to the influent sewage quantity constant. Therefore, the optimum manipulated variable is obtained by dividing the optimum manipulated variable by the influent sewage quantity. You may obtain | require as an optimal ratio fixed control which calculates | requires a ratio and multiplies this inflow sewage amount which changes every moment, and determines an operation amount. In this way, even if the process sensor is abnormal and the situation where feedback control cannot actually be performed occurs, the robustness to the above (1) and (2) is lost, but the appropriate The operation amount can be determined. This has the following merit compared to the ad hoc method of keeping (holding) the operation amount at the current value if the process sensor value is abnormal.

  In other words, if the operation amount is held when the sensor value is abnormal, the operation amount obtained by feedback control based on this sensor information when the sensor value is determined to be abnormal is considerably different from the optimal operation amount. It is highly possible that However, since the operation amount obtained by the method of the fifth embodiment is obtained from the answer obtained by solving the optimization problem, it is apparent that an appropriate operation amount is given rather than the held operation amount. become.

  On the other hand, if the process sensor abnormality diagnosis unit 14 determines that the process sensor state is FB and is normal, the contact c side of the switch unit SW2 is turned ON to select Optimal Reference Set Points (at this time, the contact d side is OFF). As a result, the optimum target values for the actuators 112 to 119 calculated by the optimum target value calculation unit 6 are output to the local control unit 3. Based on the optimum target value and the measurement data from the process measurement data collection unit 2, the local control unit 3 performs the step inflow pump 112, blowers 113 and 114, the carbon source input pump 115, the circulation pump 116, and the return sludge pump 117. The flocculant charging pump 118 and the excess sludge extraction pump 119 are controlled. Since the contents of the control at this time have already been described in the first embodiment, only the control for the blower 114 will be briefly described below for reference.

ここで、Ｋ^b _p，Ｔ^b _I，Ｔ^b _d は、ブロワ１１４の曝気風量をＰＩＤ制御する場合におけるＰＩＤコントローラの比例ゲイン、積分時間、微分時間を表すパラメータである。また、Ｓ^ref _nh4は最適目標値演算部６により演算されたアンモニア濃度の最適目標値であり、Ｓ_nh4（t）はアンモニアセンサ１２４により計測されたアンモニア濃度の計測値である。ここで、上記のパラメータＫ^b _p，Ｔ^b _I，Ｔ^b _d は、例えばIMCチューニングといった一般に良く知られている調整法を用いて調整しておくことができる。これが、第２のローカル制御回路３２が行う制御の内容である。 That is, the second local control circuit 32 takes in the measurement data of the ammonia sensor 124 from the second data collection unit 22 and also takes in the optimum target value for the blower 114 from the optimum target value calculation unit 6 to obtain the optimum aeration air volume. Qb is obtained based on the PID calculation formula of the following formula (42).

Here, K ^b _p , T ^b _I , and T ^b _d are parameters representing the proportional gain, integration time, and differentiation time of the PID controller when the aeration air volume of the blower 114 is PID controlled. S ^ref _nh4 is the optimum target value of the ammonia concentration calculated by the optimum target value calculation unit 6, and S _nh4 (t) is the measured value of the ammonia concentration measured by the ammonia sensor 124. Here, the parameters K ^b _p , T ^b _I , and T ^b _d can be adjusted using a generally well-known adjustment method such as IMC tuning. This is the content of the control performed by the second local control circuit 32.

  According to the series of operations of the fifth embodiment as described above, optimal sewage treatment process control that is robust to sensor abnormalities on a plant-wide basis can be performed. However, the optimum target value or optimum manipulated variable in the series of operations described above should actually differ depending on the quantity and quality of the incoming sewage. In particular, the amount and quality of influent sewage usually changes due to differences in weather such as rainy weather and clear weather, differences between holidays and weekdays, differences between night and daytime, and differences in season.

  Further, since it is not known in advance at what timing the sensor abnormality occurs, it is necessary to reflect the detection of the sensor abnormality in the process optimization unit 18, and such optimization calculation needs to be repeated at a predetermined cycle. . However, depending on the state of the sensor, the result of the optimization calculation by the process optimization unit 18 may be frequently switched between when the optimal operation amount is output and when the optimal target value is output for each cycle. It can happen. If this happens, it can cause the process to become unstable. Therefore, for example, sensor abnormality diagnosis results are stored over the past several cycles, and when the abnormality continues continuously or when normality continues continuously, the optimum operation amount and the optimum target value are switched. It is preferable to keep it. By providing such a hysteresis characteristic, process instability due to unnecessary switching can be avoided.

(Effect of Embodiment 5)
According to the present embodiment, when the sensor is operating normally, the process control is made robust by feedback (+ feedforward) control. On the other hand, when the sensor becomes abnormal, the feed is controlled. By optimizing the process approximately by the optimization control by the forward, as a result, it is possible to provide an optimal process control apparatus that is robust against sensor failure.

<Embodiment 6>
(Configuration of Embodiment 6)
The sixth embodiment has the configuration shown in FIGS. 7 and 9 in substantially the same manner as the fifth embodiment. However, the configuration of the advanced sewage treatment process 1 is not shown in FIG. 8 but shown in FIG. 10 differs from FIG. 8 in that an ORP sensor 130 for measuring the oxidation-reduction potential difference and a DO sensor 131 for measuring the dissolved oxygen concentration are provided in the aerobic tank 105, and COD, TN, The point is that a plurality of water quality elements such as TP and a discharge sewage sensor 132 for measuring the amount of water are provided in the final sedimentation place 106.

  Further, in the fifth embodiment, the number of process sensors includes one or a plurality of cases. However, in the sixth embodiment, it is assumed that the number of process sensors is plural. And when a certain sensor is abnormal, it can switch to the control based on the measurement of another sensor. That is, in this embodiment, an operation amount for a certain actuator can be selected from a plurality of types of target values according to a predetermined priority. Therefore, in this embodiment, it is necessary not only to determine whether an abnormality has occurred in any of the process sensors, but also to identify which sensor has an abnormality. Therefore, in the sixth embodiment, the method such as the principal component analysis described in the fifth embodiment is not sufficient, and the process sensor abnormality diagnosis unit 14 uses each of the sensors described in Table 3 above. It is necessary to judge normality or abnormalities.

  Hereinafter, the control performed by the second local control circuit 32 will be described as an example of the control of the sixth embodiment.

  FIG. 11 is a block diagram showing a configuration of the second local control circuit 32. As shown in this figure, the second local control circuit 32 has first to fourth controllers 321 to 324.

  The first controller 321 can control the aeration air volume of the blower 114 based on the NH 4 optimum target value given from the optimum target value calculation unit 6 and the NH 4 measurement value from the ammonia sensor 124.

  The second controller 322 can control the aeration air volume of the blower 114 based on the DO optimum target value given from the optimum target value calculation unit 6 and the DO measurement value from the DO sensor 131.

  The third controller 323 can control the aeration air volume of the blower 114 based on the ORP optimum target value given from the optimum target value calculation unit 6 and the ORP measurement value from the ORP sensor 130. Since the ORP value is not included in the normal process model, it is incorporated into the process model by, for example, examining the correlation with the DO concentration or ammonia concentration in advance.

  The fourth controller 324 includes the optimum target value of the discharge TN-oxygen tank NO 3 given from the optimum target value calculation unit 6, the discharge TN measurement value from the discharge sewage sensor 132, and the anox tank NO 3 measurement value from the nitric acid sensor 122. The aeration air volume of the blower 114 can be controlled based on the difference between the blower 114 and the blower 114.

(Operation of Embodiment 6)
When the aeration air volume control of the blower 114 is performed, the aeration air volume control using the ammonia sensor is most preferable if the ammonia sensor is normal. On the other hand, since the discharge TN is considered to mainly contain ammonia and nitric acid, the discharge TN-reactor end tank NO3 can control the aeration air volume considering that it is almost proportional to the ammonia concentration. It is not reliable. Therefore, the priority order of the controllers used for the aeration air volume control is set in the order of the controllers 321 to 324.

  Therefore, unless the process sensor abnormality diagnosis unit 14 diagnoses the ammonia sensor 124 as abnormal, in the process optimization unit 18, the optimum target value calculation unit 6 gives the optimum target value of the ammonia concentration to the first controller 321. At this time, a flag indicating that a target value is not given (for example, an impossible numerical value such as “10000” is set) is set for the other controllers 322 to 324.

  On the other hand, when the process sensor abnormality diagnosis unit 14 diagnoses that the ammonia sensor 124 is abnormal, the process optimization unit 18 receives the diagnosis result and the optimum target value calculation unit 6 receives the second priority with the second priority. The optimum target value of the DO concentration is given to the controller 322 of FIG. At this time, a flag indicating that the target value is not given is set for the other controllers 321, 323, and 324 in the same manner as described above.

  When the process sensor abnormality diagnosis unit 14 diagnoses that the DO sensor 131 is also abnormal, the optimum target value calculation unit 6 similarly applies the third target 323 having the third priority to the optimum target of ORP. In addition, when it is diagnosed that the ORP sensor 130 also has an abnormality, the fourth target controller 324 of the last priority is given the discharge TN-oxygen tank NO3 optimum target value.

  When the process sensor abnormality diagnosis unit 14 diagnoses that the discharge sewage sensor 132 and the nitric acid sensor 122 are also abnormal, that is, when it is diagnosed that all the sensors related to the aeration flow control of the blower 114 are abnormal, The sensor abnormality diagnosis unit 14 turns off the contact c of the switch unit SW2 and turns on the contact d. Thereby, the supply of each optimum target value from the optimum target value calculation unit 6 to the controllers 321 to 324 is stopped, and instead, as described in the fifth embodiment, the optimum operation amount from the optimum operation amount calculation unit 19 Is directly output to the blower 114.

The switching of each controller and the switching to the optimum operation amount when controlling the blower 114 described above are performed based on the solutions of the following five optimization problems.

Optimal Reference Set Points1 = arg NO3-N, NH4-N, MLSS, PO4-P, A-SRT min Cost function subject to (Mass balance, Actuator limit, Output Limit)}

Optimal Reference Set Points2 = arg NO3-N, DO, MLSS, PO4-P, A-SRT min Cost function subject to (Mass balance, Actuator limit, Output Limit)}

Optimal Reference Set Points3 = arg NO3-N, ORP, MLSS, PO4-P, A-SRT min Cost function subject to (Mass balance, Actuator limit, Output Limit)}

Optimal Reference Set Points4 = arg NO3-N, TN-NO3, MLSS, PO4-P, A-SRT min Cost function subject to (Mass balance, Actuator limit, Output Limit)}

Optimal Reference Set Points5 = arg Qstep, Qb, Qpac, Qcb, Qcirc, Qret, Qwaste min Cost function subject to (Mass balance, Actuator limit, Output Limit)}

The controllers 321 to 324 calculate the aeration air volume Qb (t) based on the following equations (43) to (46). ^Kb * _p , ^Tb * _I , and ^Tb * _d (*: 1 to 4) in these equations are parameters representing the proportional gain, integration time, and derivative time of PID control. Further, the four symbols with the superscript ref indicate the optimum target value given from the optimum target value calculation unit 6. Of these four target values, only one of them contains the correct value, and the other three target values are flags containing the above-mentioned impossible numerical values.

(Effect of Embodiment 6)
When there are multiple candidates as process values required to control a certain manipulated variable, we provide a plant-wide optimal process control device that is robust against sensor failure by preparing multiple feedback controls with each sensor. it can.

<Embodiment 7>
(Configuration of Embodiment 7)
FIG. 12 is a configuration diagram of the seventh embodiment. The seventh embodiment has substantially the same configuration as that of the fifth embodiment, except that a process sensor abnormality having a process fluctuation detecting unit 141 instead of the process sensor abnormality diagnosis unit 14 of FIG. This is a point using the diagnosis unit 14A.

(Operation of Embodiment 7)
The case of the ammonia sensor 124 will be described as an example. For example, the allowable fluctuation range of the ammonia sensor 124 is 0 [mg / L] to 4 [mg / L], and the allowable maximum change rate of fluctuation is 1 [mg / L / Set in advance as [10 minutes]. The process variation detecting means 141 determines that “there is a process variation” when such an allowable value is exceeded.

  Then, the process sensor abnormality diagnosis unit 14A diagnoses that “the sensor is normal”, and if the process variation detection unit 141 determines that “there is process variation” at that time, the embodiment The flag FB as described in 5 is set. If not, that is, if the process sensor abnormality diagnosis unit 14A has diagnosed that “the sensor is abnormal” or the process variation detection means 141 determines that “there is no process variation”, the flag FF is set. Make sure that The operation after the flag FB or the flag FF is set is the same as that in the fifth embodiment.

  The difference between the seventh embodiment and the fifth embodiment is the condition for setting the flag FB. In the fifth embodiment, if the process sensor is normal, the flag FB is uniformly set. However, in the seventh embodiment, only when it is determined that the process sensor is normal and there is process variation, the flag FB is set. In this way, feedback control is performed.

  This makes it possible to avoid unnecessarily moving the manipulated variable when, for example, the control purpose does not want to keep the process value constant, but wants to keep it within a certain range. May reduce the cost associated with the amount of operation.

(Effect of Embodiment 7)
When the control purpose is not to control the process sensor value to a certain value, but to control with a constraint that wants to keep it within a certain range, avoid moving the manipulated variable unnecessarily, and Control that minimizes the evaluation value can be performed, and cost reduction can be achieved. Furthermore, when there is a process variation, robustness can be added by executing feedback control.

The figure which shows the structural example and arrangement | positioning of a process sensor of the sewage advanced treatment process as an example to which this invention is applied. The block diagram which shows the structural example of Embodiment 1 of the plant wide optimal control apparatus by this invention applied to the sewage advanced treatment process of FIG. The figure which shows the structural example and arrangement | positioning of a process sensor of an advanced sewage treatment process as an example which this invention applies in relation to Embodiment 2. FIG. The block diagram which shows the structural example of Embodiment 2 of the plant wide optimal control apparatus by this invention. Explanatory drawing for demonstrating Embodiment 3 of the plant wide optimal control apparatus by this invention. Explanatory drawing for demonstrating Embodiment 4 of the plant wide optimal control apparatus by this invention. The block diagram of Embodiment 5 of the plant wide optimal control apparatus by this invention. Explanatory drawing which shows the detailed structure of the sewer advanced treatment process 1 in FIG. Explanatory drawing which shows the detailed structure of the process measurement data collection part 2 and the local control part 3 in FIG. Explanatory drawing which shows the detailed structure of the sewer advanced treatment process 1 in Embodiment 6 of this invention. The block diagram which shows the structure of the 2nd local control circuit 32 in Embodiment 6 of this invention. The block diagram of Embodiment 7 of the plant wide optimal control apparatus by this invention. Explanatory drawing explaining the positioning of this invention in relation to various control systems.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Sewage advanced treatment process 101 First sedimentation tank 102 1st anaerobic tank 103 2nd aerobic tank 104 3rd anaerobic tank 105 4th aerobic tank 106 Final sedimentation tank 111 Excess sludge extraction pump 112 Step inflow pump 113 Blower 114 Blower 115 Carbon source charging pump 116 Circulation pump 117 Return sludge pump 118 Flocculant charging pump 119 Excess sludge extraction pump 120 Alkaline agent charging pump 121 Ammonia sensor 122 Nitric acid sensor 123 MLSS (active sludge amount) sensor 124 Ammonia sensor 125 Phosphoric acid sensor 126 SS sensor 127 Flow sensor 128 Sewage inflow sensor 129 Alkalinity / pH sensor 130 ORP sensor 131 DO sensor 132 Outflow sewage sensor 2 Process measurement data collection units 21 to 28 First to eighth data collection units 3 Low Le control unit 31 first local control circuit (step inflow pump control circuit)
32 Second local control circuit (blower control circuit)
33. Third local control circuit (carbon source input pump control circuit)
34 Fourth local control circuit (circulation pump control circuit)
35 Fifth local control circuit (return sludge pump control circuit)
36. Sixth local control circuit (flocculating agent charging pump control circuit)
37 Seventh local control circuit (excess sludge extraction pump control circuit)
4 Optimal evaluation function setting unit 5 Constraint condition setting unit 51 Storage amount constraint setting unit 52 Output constraint setting unit 53 Operation amount constraint setting unit 6 Optimal target value calculation unit 7 Process abnormality determination unit 8 Process abnormality notification unit 9 Process abnormality diagnosis unit 10 Process abnormality control unit 11 Manual target value setting unit 12 Manual target value determination unit 13 Operator 14 Process sensor abnormality diagnosis unit 15 External input measurement data collection unit 16 External input sensor abnormality diagnosis unit 17 Manual analysis data storage unit 18 Process optimization Unit 19 Optimal operation amount calculation unit SW1, SW2 switch unit

Claims (29)

A plant-wide optimal process for controlling a process having at least one process sensor capable of measuring the state of the target process, at least one actuator capable of changing the state of the target process, and an external input to the target process In the control device,
A local control unit that operates the actuator based on a process measurement value obtained by the process sensor and compensates for dynamic fluctuations of the process so as to have desirable characteristics, and a substance amount and an energy amount of at least an external input of the process Set by a constraint condition setting unit that sets static constraint conditions for the conservation rule of storage amount, an optimum evaluation function setting unit that sets an evaluation function that evaluates operating costs and multiple control performances, and the constraint condition setting unit And an optimum target value calculation unit that supplies an optimum target value of the local control unit based on the evaluation function set by the optimum evaluation function setting unit. Plant-wide optimal process control device.   As a result of controlling the process by the local control unit given the optimal target value calculated by the optimal target value calculation unit, when the controlled variable derived from the process sensor cannot achieve the optimal target value, 2. The plant-wide optimum process control device according to claim 1, further comprising a process abnormality notification device for notifying a process abnormality.   As a result of controlling the process by the local control unit given the optimal target value calculated by the optimal target value calculation unit, when the controlled variable derived from the process sensor cannot achieve the optimal target value, 2. A plant-wide optimum process control according to claim 1, further comprising a process abnormality diagnosis device that judges a process as abnormal, automatically investigates the cause of the abnormality, and reports the investigation result to a process manager. apparatus.   When the controlled amount derived from the process sensor cannot achieve the optimal target value as a result of the local control unit that is given the optimal target value calculated by the optimal target value calculating unit controlling the process, A process abnormality control device incorporating a mechanism for judging a process as abnormal, automatically investigating the cause of the abnormality, and automatically recovering the process based on the investigation result. The plant-wide optimum process control device according to 1.   2. The plant-wide optimum process control device according to claim 1, further comprising a manual target value calculation unit that can be switched to and from the optimum target value calculation unit.   2. The plant-wide optimum process control apparatus according to claim 1, wherein a constrained nonlinear programming method is applied to the optimum target value calculation unit.   2. The plant-wide optimum process control device according to claim 1, wherein a semi-optimization method using metaheuristics is applied to the optimum target value calculation unit.   Static constraint conditions related to conservation laws such as material balance and energy balance of the constraint condition setting unit are updated at a predetermined cycle according to a change in external input, and in synchronization with this, calculation in the optimum target value calculation unit The plant-wide optimum process control apparatus according to claim 1, wherein:   The plant-wide optimum process according to claim 1, wherein an external input related to a static constraint condition relating to a conservation law such as a material balance and an energy balance of the constraint condition setting unit is constructed from only measurable variables. Control device.   Estimating that the external input related to the static constraints related to conservation laws such as the material balance and energy balance of the constraint setting section consists of measurable variables and unmeasurable variables, and estimates the values of unmeasurable variables The plant-wide optimum process control apparatus according to claim 1, further comprising means.   The plant-wide optimum process control device according to claim 1, wherein a parameter of a control law of the local control unit is changed according to the optimum target value supplied from the optimum target value calculation unit.   The plant-wide optimum process control apparatus according to claim 1, wherein the local control unit has a control law using a dynamic model of the process.   The plant-wide optimum according to claim 1, further comprising: a display device that displays an output result of the optimum target value setting unit, wherein the process controller / operator is manually operated by the process controller / operator. Process control device.   2. The process according to claim 1, wherein the process is a sewage treatment process, and the optimum evaluation function setting unit uses a cost evaluation function that simultaneously considers the discharge water quality cost obtained by converting the discharge water quality and the operation cost related to operation. The plant-wide optimum sewage treatment process control device described.   The process is a sewage treatment process, and as the constraint conditions, regarding inflow sewage as an external input, a variable that can be converted into COD concentration, a variable that can be converted into TOC, a variable that can be converted into BOD, and an ammonia concentration Variable that can be converted to nitric acid concentration, variable that can be converted to Kjeldahl nitrogen concentration, variable that can be converted to total nitrogen concentration, variable that can be converted to phosphoric acid concentration, and total phosphorus concentration A variable that can be converted, a variable that can be converted to MLSS concentration, a variable that can be converted to SS concentration, a variable that can be converted to pH, a variable that can be converted to ORP, and a variable that can be converted to alkalinity 2. The plant-wide optimum process control apparatus according to claim 1, wherein the substance quantity conservation law is described by an algebraic equation using at least one of the variables.   In addition to the variables of the constraint conditions, as an external input, using a variable representing a plurality of elements of organic matter measured by COD and a plurality of elements of sludge containing microorganisms measured by MLSS, The plant-wide according to claim 15, further comprising: an estimation unit that estimates a plurality of elements of an organic substance described by an algebraic equation and measured by the COD and a plurality of elements of sludge measured by the MLSS. Optimal sewage treatment process control device.   The process is a sewage treatment process, and the local control unit calculates the aeration air volume, which is one of the objects to be controlled by the actuator, from a DO meter, an ORP meter, an ammonia meter, a nitric acid meter, and a TN meter as the process sensors. 2. The plant-wide optimum process control device according to claim 1, wherein the control is performed using a detection output of at least one of the sensors.   The process is a sewage treatment process, and the local control unit determines an amount of returned sludge, which is one of the objects to be controlled by the actuator, as an ammonia meter, a TN meter, a TP meter, a nitric acid meter, and a phosphoric acid as the process sensors. 2. A plant-wide optimum process control apparatus according to claim 1, wherein control is performed using a detection output of at least one of a meter, an ORP meter, an MLSS meter, a UV meter, a COD meter, a BOD meter, and a TOC meter. .   The process is a sewage treatment process, and the local control unit extracts a surplus sludge extraction amount, which is one of the objects to be controlled by the actuator, with a phosphate meter, a nitrate meter, an ammonia meter, and a TP meter as the process sensors. 2. The plant-wide optimum process control device according to claim 1, wherein control is performed using a detection output of at least one sensor of the TN meter and the MLSS meter.   The process is a sewage treatment process, and the local control unit determines a circulation amount, which is one of objects to be controlled by the actuator, as a process sensor, a nitric acid meter, an ORP meter, a TOC meter, a COD meter, and a BOD meter. 2. The plant-wide optimum process control device according to claim 1, wherein the control is performed by using a detection output of at least one of the UV meter and the ammonia meter.   The process is a sewage treatment process, and the local control unit calculates a carbon source input amount that is one of the objects to be controlled by the actuator by using a nitrate sensor, a TN meter, an ORP meter, a COD meter, and a UV sensor as the process sensors. 2. The plant-wide optimum process control device according to claim 1, wherein the control is performed using a detection output of at least one sensor among a meter, a BOD meter, and a TOC meter.   The process is a sewage treatment process, and the local control unit calculates a carbon source input amount, a circulation amount, and a step inflow amount, which are one of the objects to be controlled by the actuator, from a COD meter and a UV meter as the process sensors Using the detection outputs of at least two sensors in the TOC meter, BOD meter, nitric acid meter, TN meter, and ORP meter, simultaneously control the organic substance amount and nitric acid amount ratio to a predetermined value. The plant-wide optimum process control device according to claim 1.   The process is a sewage treatment process, and the local control unit calculates a flocculant input amount, which is one of the objects to be controlled by the actuator, among a phosphate meter, a total phosphorus meter, and an ORP meter as the process sensors. 2. The plant-wide optimum process control apparatus according to claim 1, wherein the control is performed using a detection output of at least one sensor.   The process is a sewage treatment process, and the local control unit is one of the objects to be controlled by the actuator, and the step inflow is measured by using a phosphate meter, a TP meter, a nitrate meter, a TN meter, and an ORP meter as the process sensors. 2. The plant-wide optimum process control apparatus according to claim 1, wherein control is performed using a detection output of at least one of a sensor, a COD meter, a UV meter, a BOD meter, and a TOC meter. At least one actuator capable of changing the state of the target process;
At least one process sensor that can measure several states of the target process and includes a process sensor used to determine an operating amount of the actuator;
An external input (disturbance) to the target process and at least one external input sensor for measuring this,
In any process having
A process sensor abnormality diagnosis unit for determining whether the process sensor is in a normal state or an abnormal state;
An external input sensor abnormality diagnosis unit for determining whether the external input sensor is in a normal state or an abnormal state;
A local controller that operates at least one of the actuators to compensate for dynamic variations in the process to have desirable characteristics based on process measurements obtained by the at least one process sensor;
When the external input sensor is normal in the external input sensor abnormality diagnosis unit, this external input sensor value can be input, and in the case of abnormality, an external input value analyzed by manual analysis or the like can be input. A constraint condition setting unit having, as at least one constraint condition, a static constraint condition related to a conservation law of a conservation amount such as the amount of substance and energy,
An optimal evaluation function setting unit that can set an evaluation function that can evaluate the operating cost and multiple control performances;
When the constraint condition set by the constraint condition setting unit is input and the output of the process sensor abnormality diagnosis unit is normal based on the evaluation function set by the optimum evaluation function setting unit, the local control unit A process optimization unit that can output an optimal operation amount to the actuator when the output of the process sensor abnormality diagnosis unit is abnormal.
A plant-wide optimum process control device characterized by comprising: At least one actuator capable of changing the state of the target process;
At least two process sensors, including at least two process sensors capable of measuring several states of the target process and used to determine an operating amount of the actuator;
An external input (disturbance) to the target process and at least one external input sensor for measuring this,
In any process having
A process sensor abnormality diagnosis unit that determines whether each of the plurality of process sensors is in a normal state or an abnormal state;
An external input sensor abnormality diagnosis unit for determining whether the external input sensor is in a normal state or an abnormal state;
Locally actuates at least one of the actuators based on any process measurement by at least two of the prioritized process sensors to compensate for dynamic variations in the process with desirable characteristics A control unit;
When the external input sensor is normal in the external input sensor abnormality diagnosis unit, this external input sensor value can be input, and in the case of abnormality, an external input value analyzed by manual analysis or the like can be input. A constraint condition setting unit having, as at least one constraint condition, a static constraint condition related to a conservation law of a conservation amount such as the amount of substance and energy,
An optimal evaluation function setting unit that can set an evaluation function that can evaluate the operating cost and multiple control performances;
The constraint condition set by the constraint condition setting unit is input, and based on the evaluation function set by the optimal evaluation function setting unit, an optimal target value calculation unit that outputs an optimal target value to the local control unit, and at least A process optimization unit having an optimum operation amount calculation unit that outputs an optimum operation amount to one or more actuators;
With
In addition, the optimum target value calculation unit selects any one of the two or more process sensors in order from the normal sensor and the one with the highest priority, and the optimum for the control using the selected sensor. A target value is output to the local control unit, and the optimal operation amount calculation unit is configured to perform an optimal operation on the at least one actuator when all of the two or more process sensors are abnormal. Output the quantity,
A plant-wide optimum process control device characterized by that. At least one of the process sensor abnormality diagnosis unit and the external input sensor abnormality diagnosis unit is:
Use statistical methods such as principal component analysis based on correlation by multiple sensors,
27. The plant-wide optimum process control apparatus according to claim 25 or claim 26. The process optimization apparatus normally outputs an optimum operation amount as feedforward control, and when the process state value deviates from a fluctuation range determined in advance by the process sensor, the process sensor abnormality diagnosis unit performs a process sensor. Is not abnormal, and if it is not abnormal, it switches to feedback control by outputting the optimal target value.
27. The plant-wide optimum process control apparatus according to claim 25 or claim 26. In switching between the optimum target value output and the optimum manipulated variable output in the process optimization apparatus, a bidirectional changeover switch having a hysteresis characteristic is provided.
27. The plant-wide optimum process control apparatus according to claim 25 or claim 26.

Images